4th Meeting - 2nd Session Group of Governmental Experts on Lethal Autonomous Weapons Systems 2025 Meetings & Events Date: 2 September 2025 Language: English Transcript: https://transcripts.un.org/es/asset/k19/k197usei81?lang=en Transcripts available through this tool are created by using automatic speech recognition and are not official records nor official documents of the United Nations. Official records and official documents are available on the Official Document System of the United Nations. --- Chair [0:01]: Good afternoon, excellencies, distinguished delegates. Many thanks for being back in the room. That's helpful. I already had a request for the floor from the US. Please, they would like to still continue on box 2, so that will be no problem since Probably the issue they are going to address will also be in box 3. And then we move to box 3. Let me now pass the floor to Distinguished Representative of the US. Sir, you have the floor. United States of America [0:46]: Thank you very much, Chair, for giving us the floor. And apologies for coming in after lunch with additional reflections on the box 2 discussion. We've listened carefully to the discussion on paragraph five, which in many ways is a continuation of the good discussion that we had in March on your proposal to include context appropriate human control and judgment in the rolling text. While we recognize that you've made some adjustments to the language that we looked at in March, for example, to indicate that context appropriate human judgment and control are needed to ensure compliance with IHL, Nearly every delegation that supports this phrase has explained that their position is that some form of human control is or should be required by IHL. We continue to hear delegations say that this reflects an existing IHL requirement. We continue to hear delegations say that it is a prerequisite or essential to compliance with international humanitarian law, which in our view is tantamount to saying that it is an existing requirement. And this is simply wrong. It's simply wrong as a matter of law. Yet it is also the case that almost every delegation seems to acknowledge that there is not yet a common understanding of context-appropriate human judgment and control, as was the case in March We heard a number of delegations say that human control is required, but it doesn't actually mean that the human must control the weapon during its operation, or that there are other forms or means of indirect control that could be understood as human control that would be required to comply with IHL. This leads us to two observations, Chair. First, if the group negotiates a formula of words and the first thing we have to say when we're explaining that formula of words to others is that it doesn't actually mean what it sounds like it means, then from my delegation's perspective, we've chosen the wrong words. As we said in March, we're negotiating language about what international law requires. The ordinary meaning of the words we choose matters. If we say one thing and mean another, we risk confusing and not clarifying legal requirements. Control, from our perspective, is different from judgment. Control and judgment also are necessary for compliance with IHL rules, but they're not sufficient. And that's one of our key concerns, Chair, Those that are advocating for this formula of words are explaining that if you do these six words, that is sufficient for compliance, which we just simply don't accept. IHL violations are committed with poor human judgment and full human control over the weapon system. That's just the reality. Second, it seems very clear to us that we're still at a point in this debate where there is no shared understanding of what is meant by these six words. In that context, it's difficult for our delegation to accept, as some delegations have asserted, that this reflects an organizing principle that helps us move forward. And it is impossible for our delegation to accept that we've hit on a form of words that codify existing IHL requirements. Delegations are being asked to accept a general formulation that is intended to reflect a prerequisite for compliance with IHL now, with the recognition, even by those who are advocating for it, that there is no shared understanding of what these terms mean, and that we should develop those understandings later, perhaps in its own box or in a technical annex. that could include, as suggested by the delegation of Ukraine, a detailed definition of each of these six terms. Chair, we continue to think that this approach has it backwards. The delegation of Pakistan said yesterday that in their assessment, the purpose of establishing the principle of control is to help us understand those measures that are needed to set limits or parameters to ensure that the weapons will operate in a manner consistent with IHL. I think I took very careful notes of that statement. And as the delegate from Norway explained, those supporting this framing are seeking to explain what is required to ensure compliance with IHL. We 100% agree, 100% agree, that the purpose of this GGE is to identify what applicable IHL rules require and what measures are needed to meet those requirements. including relevant limits or parameters to ensure that the weapon will be used consistent with IHL. But the requirements of all of IHL and the measures that can be taken by states to promote IHL compliance cannot be reduced to a set of five or six words, which give no indication of what control or what judgment is actually required. As the interventions of Ukraine and Armenia helpfully highlighted, we need to be specific about what is actually required, rather than just generally saying that controller judgment is required. So, Chair, from our delegation's perspective, a better approach would be to specifically describe those measures that are needed for the use of laws to be consistent with specific IHL rules in the elements of this instrument. In our view, this is the work to be done in boxes three, four, and five. The purposes of boxes four and five, in our view, is to identify those measures that are needed to ensure that laws are used in compliance with IHL requirements articulated in box three. In our proposals, we've recognized that IHL imposes specific requirements and proposed particular measures that, given the nature of law, should be implemented to ensure that laws are used lawfully. So let's do that. Let's develop specific understandings of what is required and what measures can help fulfill those requirements. Thank you, Chair. Chair [7:13]: I thank the distinguished representative of the U.S. for his statement regarding his delegation's position on box 2, and in particular para 5. We have, of course, again, context-appropriate human judgment and control coming back, as he also referred to in box 3, and it's also in box 4. In box 3, it's indeed followed by a number of measures. So And that is what we have on our plate this afternoon, box three. So I think this has been an excellent bridge to the discussion we are going to have now. If I look around the room, I recognize Austria asking for the floor. I would also invite other delegations to put up their plaques. And if you're in the back row, please wave your plaque. But now let's start with the distinguished representative of Austria. So you have the floor. Austria [8:19]: I think we had a very similar exchange with the same arguments already in our last session, but still we think it's an important conversation to have because we don't want to have any open questions when moving forward. And in this regard, it's important that we address also all these maybe shortcomings or problems that we have to solve when we move forward and that might also raise questions outside of this room with those people who have not been part of our conversation in the future. So we definitely have to look into all these different problems. I just have three points, or maybe four, on what has been mentioned, and I think many of them have been already made in the conversation this morning. I think number one, and this is, I think, maybe the most important one, the reason why we are using different language for these kind of weapons, and I think Norway made this point, is that these weapon systems are different from other weapon systems that we are seeing under IHL and that we tried to regulate or cover so far. The autonomy is a game changer and it just makes them to some extent different and we need a way to translate existing IHL so that it also applies to those weapon systems. It does not mean that we want to make them, anthropomorphize them in some way. The link has to be over the human user. And the human control is just the way how we move from existing IHL to the weapon systems in place. And to some extent, this is just different from existing weapon systems because the autonomy is taking over some quite important functions when it comes to implementation of IHL. So this is a huge difference from previous previous weapon systems or previous applications of IHL that we have seen so far, so this is why we probably have to use new words, new concepts that are a bridge of translation, but also in general, I think it wouldn't be out of the ordinary to use new concepts in the face of, of new challenges. The second point is that arguments that there is this and that concept that does not exist yet, I think this would be an argument that you could use for all multilateral negotiations, would it be the plastic treaty or whatever comes onto our plate, there will be always Some words that have never been used before. I think Additional Protocol I probably had a lot of substance that we didn't see anywhere before. Some treaties built themselves on other documents that have been previously there. But often there is a number of concepts or even ideas that are a novelty. And I think we are in one of these very, very important processes where something like this might also be needed. The third point, and this is maybe the good message, I think that we agree with each other that we need to go into specificities and that we should not develop hollow concepts or hollow ideas. We have to fill them with very specific ideas and we also have to explain how they can be achieved. And this is what we do in the next box, where we actually set out regulations or specific measures that states have or could or should undertake in order to achieve human control, context appropriate human judgment and control, and just something that they have to undertake to actually comply with whatever we set out in this text or as possible elements of an instrument. This is no easy task and we have to move forward and these are just a few comments that we wanted to make. And yeah, in general, we need some specific explanation on how IHL is implemented in the right way. Maybe there is a way also to do it without translation concepts, but it just makes our lives much more easier, even if there is some kind of unclarity at the same time, it also is something that the human mind can grasp and something that actually pushes us forward in our discussion. So, I mean, this has been very philosophical, what I mentioned, but I think This is just the way how multilateral discussions work in general. Thank you. Chair [12:35]: I thank the distinguished representative of Austria. Next on my list is the distinguished representative of the Russian Federation. Sir, you have the floor. Russian Federation [12:49]: Thank you very much, distinguished Chair. We're obliged to take the floor once again. in light of the fact that the comments that were made by colleagues from the United States of America and Austria, those comments are in fact quite important. And in many ways, they will be predetermining our future discussion on very important issues, including the question of control. We believe that we shouldn't really try to convince each other on this question. We understand very clearly that each delegation has their national position, and Those positions were presented in great detail, and that includes the delegation of the United States and Austria and other States Parties. We believe that we should focus on how can we bring these national positions closer to one another and to be able to find compromise-based, consensus-based understandings of this question. Now, in light of what we heard, we do not believe that this problem is something that cannot be resolved and that the positions that were presented here are somehow counter to one another and cannot be reconciled. We see the possibility to find a very specific and common understanding on this question. And we believe that that can be done step by step. First, we need to take a look at the different options that we may find in connection with the ideas contained in paragraph 5 of box number 2, namely what indeed is a factor that will ensure compliance with international law, including international humanitarian law, in particular the principles that have been identified in IHL. And here we can be helped by the guiding principles that have been already approved by states parties, in particular, guiding principle C. And that is indeed the objective of guiding principle C. Guiding principle C talks about exactly the same thing. How can we ensure compliance with IHL? And in this context, we'd like to recall that this is not simply some understanding that was adopted and approved by our group. It is a guiding principle, guiding principles that have been adopted by all States Parties to the Convention. And this indeed is a relevant decision that was taken at a political level, a decision taken by states. And this is why we can say that at least we have this principle, this guiding principle that could be used in order to implement the mandate of our group. as far as the notion of context appropriate is concerned. And here we also see that we have serious problems in light of the discussion that we have heard. Here, as we said earlier, we believe that a possible solution is a discussion on paragraph six of box three and perhaps other paragraphs in boxes four and five that relate to this notion, perhaps indirectly, sometimes indirectly, but in one way or another, these notions that we find in subsequent paragraphs, they also relate to the operational context that must be taken into account in order to ensure due levels of control in connection with such weapons systems. And this is why we believe that we should focus on trying to find these options to find a solution. But what we clearly understand at this stage, distinguished Chairperson, is that we hardly should focus our efforts, our attention on trying to find appropriate language in terms of identifying this or that notion. We have to focus on what the different notion actually means, what it contains, what this or that notion actually represents. And we should not try through using this or that wording to present a situation that would be not really clear, something that would be a solution for delegations. We already understand right now that this will hardly represent a solution to the problem. On the contrary, this may confuse things even further. So let us focus on the real substantive aspects. Let's discuss them. And let us try not to simply repeat national positions, but let us try to identify those possible solutions that would enjoy support of all States Parties and that could contribute to finding decisions that will be consensus-based. Thank you. Chair [19:46]: I thank the distinguished representative of the Russian Federation for stating his position and also reacting to the remarks by previous delegations. So far, no further request for the floor. So this means that box 3 is agreeable to all of you? I think you have a few lines somewhere in your notebook on that. I recognize the distinguished representative of Ukraine. Sir, you have the floor, and thanks for keeping the discussion going. Ukraine [20:24]: Thank you, Chair. The delegation of Ukraine would like to clarify whether we are now talking about box. 3 and we are ready to conclude our discussion on box 3, or we are still within box 2. Speaker 9 [20:36]: Yes. Chair [20:37]: No, we have moved to box 3 now. We have moved to box 3. Ukraine [20:41]: Thank you for the clarification, Chair. That's all. Chair [21:04]: I see Russian Federation. You have the floor. Russian Federation [21:10]: Again. Thank you very much, distinguished Chair. On box 3, for the time being, we would prefer to limit our comments to paragraphs 1 through 5. And we'll leave it, leave paragraph six as something, as a dessert. Because we believe that it requires a more serious and focused discussion that will relate to this paragraph, especially in light of the discussion on box and more specifically, paragraph 5 from box 2. So we would be ready to come back to paragraph 6 at a later stage, but I would like to refer now to paragraphs 1 through 5. First of all, a general comment, something that we already conveyed during the previous sessions of the GGE, and we would like to repeat that once again. We do not believe that it is necessary for the purposes of this group, in line with its mandate, the mandate that this group has, to focus stylistically in paragraphs 1 through 5 on language that we will be prohibitive. In other words, the inclusion of strict language like "it is prohibited" in the text. We believe that any possible prohibitions that we find in paragraphs 1 through 5, and if we are more precise, in paragraphs 1 through 4, can be representative without the use of the word "prohibited". Moreover, we managed to do that during the previous work of the group, and the relevant language has been reflected in the reports of the group in previous stages without the use of this word. This is why, for example, in paragraph 1, we would prefer to replace or delete the first part of the sentence, "It is prohibited in all circumstances to use," and instead of that, to add at the end, "the language that we used earlier, and that is a language that we agreed to and in the framework of this format, and the language must not be used." In the same way, in paragraph 3, instead of the part of the sentence that we have in the beginning, "It is prohibited in all circumstances to make," to use the verb "should not be" before the ending of the paragraph, before the words "the object of an attack by laws" and also add "as well as" before the words "individual civilians or civilian objects". In this case, the sense of paragraphs 1 and 3 will not be modified in any way. The idea that is contained in these paragraphs will be exactly the same. But in this way, we'll be reproducing the language that we have already agreed to on the basis of consensus. On paragraph 4, we believe that this paragraph is fairly limited and selective. It presents ideas in a more narrow manner than other notions that we have agreed to in this group earlier. We believe that this is something that is not acceptable. And what we have in mind is paragraph 23A from the report of the group from 2019 that does not limit things to this stage of use, but also relates to the other stages of the life cycle. In other words, the stages of development, research, testing, design, deployment, in connection with minimizing the risks that these weapon systems may have in the context of compliance with obligations under IHL. With that, paragraph 23A from the 2019 report also talks about other types of risks that must be considered by states parties. when they use or develop such weapon systems. As you see, the relevant paragraph from the 2019 report is much broader than the paragraph 4 that we see here in box 3. And we continue to be in favor of reproducing paragraph 23A from the 2019 report here, due to the fact that it fully covers from the point of view of what is contained in paragraph 4. But with all of that, it contains some very important and substantive understandings related to the broad range of the different risks that may be related to the different stages of the life cycle of such weapon systems. On paragraph 5, we believe that it is superfluous in light of the proposal that was just made by our delegation on paragraph 4, in connection to the use of paragraph 23, and as we said, paragraph 23 relates to the stage of design and development, testing, deployment of such system. And as far as the stage of use is concerned, we have reference to that in paragraph one of this section, where we clearly say in which cases such weapon systems must not be used. And this is also an idea that is reflected in paragraph 2 of this section that specifies how such systems should be employed or used. And moreover, it is also something that is covered by paragraph 3 of this box when we talk about a specific case of the use of such weapon systems, namely that such weapon systems should not be used against civilians individual civilians or civilian objects. In this connection, we suggest that we delete paragraph 5 because the ideas contained in paragraph 5 are already covered by paragraphs 1 through 4. And once again, I will stress that paragraph 4 will be, like we suggest, reproducing the text from paragraph 23A of the 2019 report. Thank you very much, distinguished Chair. Chair [30:58]: Let me thank the distinguished representative of the Russian Federation for sharing the views of his delegation with regard to the paragraphs 1 to 5. The floor is open again. I recognize the distinguished representative of India. Madam, you have the floor. India [31:28]: Thank you, Chair. I'll also share brief comments in respect of the first five paragraphs of box 3. These paragraphs, in our view, seem to be an attempt to list prohibitions or measures of a similar nature. We can support paras 1, 2, and 3 as drafted. We have some comments on para 4 and para 5. We are not entirely opposed to para 4, however, we do have several questions which, in our view, require further discussion. We are not aware of existing IHL obligations that inspire this para. We have questions around the terms used, such as "anticipated" and "controlled" regarding effects in attack. A related but more important question is whether we can, despite lack of sufficient understanding around these terms, frame them as a clear-cut prohibition. Our view is that we need more discussion. We have taken note of para 5 and can support this framing. This para takes into account concerns expressed on design and development of laws. At the same time, it frames them in a positive manner. However, we think that there is need for more discussion on whether this para belongs in this particular box. Our preference would be that we discuss its placement consistent with IHL as part of box 4, which we view as a listing of enabling or implementation measures instead of creating primary obligations. I thank you. Chair [33:20]: I thank the distinguished representative of India for stating the position of our delegation with regard to the paragraphs 1 to 5. We have two more requests for the floor. First, the Republic of Korea, to be followed by Spain. Let me now pass the floor to the distinguished representative of the Republic of Korea. You have the floor. Republic of Korea [34:05]: Thank you, Mr. Chair. First of all, my delegation is of the view that IHL principles primarily regulate the use of weapon systems, not their development, with the exception of Article 36 of the Additional Protocol. So while my delegation does not object to discussing the important element of development in this text, it should be addressed in the appropriate context. We therefore consider it inappropriate to discuss development under the section on prohibitions. Accordingly, we suggest changing the term "prohibitions" in paragraph 5 to "principles and requirements" to more accurately reflect this notion. Following up on the previous discussion regarding box 2, And there have been suggestions to move paragraph 5 to a separate box. While my delegation appreciates the efforts to move the paragraph out of box 2, we do not believe it is appropriate to address it in isolation in a separate box. Instead, we consider that this notion would be better situated within the framework of paragraph 6 of box 3 that we are talking about, as the Russian delegation has alluded to. Paragraph 6 outlines measures relating to control, human judgment, and operational context. In this regard, we also suggest moving paragraph 6 to box 4, as other delegation have noted, and box 4 primarily addresses provisions and legal requirements, while paragraph 6 enumerates possible measures to oppose such requirements. So it is logical to move that to box 4. We agree with the delegations that have emphasized that there is a need for novel approaches to address novel challenges. We also agree with that our task is to further elaborate how ICH applies to laws to ensure compliance. However, we do not agree that an overarching organizing principle such as context-appropriate human control and judgment is required to do this job. A more straightforward approach, as noted by the US delegation and Russian delegation, is to discuss specific measures such as those contained in paragraph 6 of box 3. To summarize, my delegation proposed merging paragraph 5 of box 2 with paragraph 6 of box 3, with appropriate adjustment of the wording, and then relocating the newly merged paragraph to box 4. Now, turning to the specific provisions within paragraph 6, the desert of box 3, my delegation will now repeat our position on the term context-appropriate human judgment constraint in the chapter 4. But regarding paragraph 6 and 6, we do not see value in limiting real-time machine learning or the ability to change emission parameters of laws. We believe such restrictions would constrain technological development and deprive the system of the flexibility which is essential for effective operation. Real-time machine learning can enhance both the precision and adaptability of those systems, thereby contributing to over military effectiveness and improve civilian protection. A rigid system, by contrast, could create greater risk in practice on the ground. So we would like to delete those restrictions in paragraph 6. Finally, we would like also to note that we consider paragraph 6D, Roman number 2, to be a necessary restriction and therefore recommend its removal. Thank you. Chair [37:48]: I thank the distinguished representative of the Republic of Korea. Next on my list is the distinguished representative of Spain. You have the floor. Spain [38:00]: Thank you very much, Chairperson. Well, first and foremost, once again, I would like to say how grateful we are to you for guiding us in these debates. We would like to make the following comments expressing my country's position. We support your proposals for paragraph 1, 2, 4, and 5, though we do have some comments on paragraph 3. and on my delegation approves of references to civilian population and civilian objects, individual civilians and civilian objects, given that they are concepts of IHL and they are covered in Additional Protocol I, we would like to make the following proposal. After the word "population," in English it would read. "It is prohibited in all circumstances to make the civilian population or to engage targets that do not constitute a military object, the object of an attack by laws. This proposal would be in line with what is said later in paragraph 6, roman numeral 4. We will be submitting this proposal in writing to you later. We would like to take the floor at a later point in order to make comments on paragraph 6. Thank you. Chair [39:27]: I thank the distinguished representative of Spain for sharing the position of her delegation and indeed we would appreciate to receive it also in writing this post change in the text. And I recognize the distinguished representative of the ICSC. Madam, you have the floor. ICRC [39:49]: Thank you, Chair. Following other delegations' approach, I will keep my comments to paragraphs 1 to 5, and we will come in later on paragraph 6. We support the inclusion of paragraphs 1 to 3 in this section, which have been recognized by the GGE in the Guiding Principles. They are a restatement of IHL rules and principles as they apply to laws and should be retained. We welcome the insertion of paragraph 4, prohibiting autonomous weapon systems whose effects cannot be anticipated and controlled, as required by IHL. We particularly welcome the reference to effects that cannot be anticipated, and we take this opportunity to emphasize that it does mirror existing IHL terminology. For example, the use of the word anticipated can be found in the principle of proportionality. It also provides a constructive bridge between different views expressed in the GGE, namely making a reference to the risks posed by unpredictable autonomous weapon systems while retaining IHL language. We would, however, recommend replacing the term "controlled" with "limited", as this more closely reflects existing IHL terminology. For example, this can be found in the prohibition against indiscriminate attacks. And it also avoids potential ambiguity over the meaning of "control" in this context. With regards to paragraph 5, we appreciate the explicit reference to both development and use in box 3. As highlighted by several delegations during the March 2025 GGE session, extending coverage to development supports compliance with IHL obligations, including the requirement to respect and ensure respect for IHL at all times, as well as the requirement that parties to an armed conflict take constant care to spare the civilian population, not only in attacks, but throughout their military operations. It also reflects the need to ensure that weapons are not developed which would likely be incompatible with IHL and provides greater clarity for developers. In the CCW context, prohibiting the development of such weapons can also be understood as promoting the purposes of the instrument by supporting the prohibition on their use. This is analogous to Amended Protocol II, Article 8, which, quote unquote, "in order to promote the purposes of the Protocol, prohibits the transfer of any mine the use of which is prohibited by the Protocol." One key gap in the current list of prohibitions is the lack of an explicit prohibition of anti-personnel autonomous weapon systems. While IHL does not categorically prohibit autonomous weapon systems that target humans directly, it's difficult to envisage realistic combat situations where the use of such weapons against people would not pose a significant risk of IHL violations. any restrictions that could theoretically be devised on the use of anti-personnel autonomous weapon systems to reduce the risk of IHL violations would likely be too complicated to be applied reliably and effectively in the wide variety of actual combat situations. In contrast, a complete prohibition on anti-personnel autonomous weapon systems would provide clarity for militaries and developers and would have a far greater impact on reducing the risks of strikes against people protected in armed conflict. Such a prohibition would also be an effective response to broader ethical and humanitarian concerns. I thank you, Chair. Chair [43:09]: And I thank the distinguished representative of the ICRC. And next on my list is Campaign to Stop Killer Robots. You have the floor. Campaign to Stop Killer Robots [43:20]: Thank you, Chair. We welcome the progress that States have made through the rolling text in their common understanding of some of the basic components of meaningful human control or context-appropriate human control and judgment. Many of the elements in boxes 3, paragraph 5 align with some of the basic parameters that we believe must be now fully developed into rules in a legal instrument in order to uphold basic legal and ethical principles and ensure responsibility and accountability. In box 3, these elements include the consideration of predictability, reliability, traceability and explainability, the importance of ethical as well as legal assessments, and the placing of limitations on the types of targets, duration, geographical scope and scale of operation of weapon systems. We recommend that all the elements in box 3 are retained in the text at this point, as a basis for States' subsequent and more detailed negotiations on a legally binding instrument. This particularly applies to the elements in paragraphs 5 and 6, with one exception that we will raise shortly. We welcome the recognition by States in the rolling text of the fact that context-appropriate human control and judgment is needed to uphold international law. The linkage between such control and fulfilling basic legal principles is important and must be preserved in States' discussions and negotiations. However, we do not believe that ensuring compliance with existing law is the sole purpose of establishing international rules to ensure meaningful human control in the use of force. Rather, enshrining such control as a principle in international law is crucial for drawing red lines to protect our humanity and for establishing clearly what we find fundamentally ethically unacceptable as an international community when it comes to states and companies pursuing increasing autonomy in weapons systems. The pursuit of increasing autonomy represents a fundamental shift in the use of force. It raises much bigger questions than IHL compliance alone. For this reason, we were disappointed to see the changes to paragraphs 5 and 6 in Box 3 that remove the concept of a general prohibition on the use of systems without meaningful human control. We also regret that the current text more explicitly restricts the purpose of box freeze elements to the goal of upholding compliance with IHL. We are also deeply concerned to see the addition of paragraph 6. Merely limiting real-time machine learning with regard to target selection and engagement functions will not be sufficient to prevent a loss of sufficient understanding and meaningful control over systems. We continue to urge States to fully consider the issues raised by anti-personnel autonomous weapons systems and how this must be dealt with. We believe that a specific prohibition on systems that target people within a wider legal instrument on autonomous weapons systems is the only ethically and legally viable response. Thank you, Chair. Chair [46:46]: I thank the distinguished representative of Stop Killer Robots. Next on my list is Panama. Adam, you have the floor. Panama [46:56]: Muchas gracias, your Excellency. Thank you very much, Chair. I'll follow the example of other speakers speaking before me. On box 3, I would like to refer to the first five paragraphs. And of course, I reserve the right to take the floor once again to make comments on the paragraphs that follow. But before starting my comments, I'd like to react first to one of the proposals on paragraph 5 of box 2, with the suggestion that we delete it or to bring it to link it with similar information in box 3. We do not believe that there is any duplication here, and we prefer to keep the reference in both boxes. Now, on the first three paragraphs of box 3, we agree with the comments that were made by the ICRC. These paragraphs, as presented here, and reflected in the guiding principles that were adopted by the GG in 2019 and also reflect the norms and principles of IHL. On paragraph 4, we support the paragraph as is presented here. And the reference to the anticipation of the effects, as the ICRC has already explained, it is simply the terminology that already exists in IHL related to the principle of proportionality. And we also agree to replace the term "controlled" with the term "limited." because this aligns itself with the existing terminology from IHL. On paragraph 5, we support the paragraph as drafted and to keep the references to both development and use, because the scope of the term development will ensure compliance with obligations under IHL, especially the obligations to respect and make sure IHL is respected in all circumstances. And this is something that we find in Article 1 of all the four Geneva Conventions from 1949, and it is also aligned with the objectives of the CCW, the convention. We see that in the preambular part, the general principle and protection of civilians against the effects of hostilities, and also the prohibition or limitation of means and methods of war that may lead to excessive injury or harm and have an impact on the environment. And also, these means are not limited. So to mention that, among other things. So these are my comments on the first five paragraphs of box 3. Thank you. Chair [50:41]: Let me thank the distinguished representative of Panama for sharing the comments of her delegation with regard to box 1 to 5. I have two more requests for the floor from Malawi and Austria. Let me first pass the floor to the distinguished representative of Malawi. You have the floor. Malawi [51:01]: Thank you, Chairperson. Apologies, we were not here in the morning, so I just have a comment also on two, on box two. On box two, we would also like to support the proposed text as it is and not move para 5. We believe the need for greater precision around the term context appropriate, human judgment and control. So we support the text to be the way it is. For box 3, We believe that systems with unpredictable and or uncontrollable behavior must be preemptively prohibited. We therefore strongly support the prohibitions are trying in this section. A specific comment on paragraph three, we support the explicit recognition of individual civilians. On paragraph five, we support the inclusion of prohibiting the design, development, and the use of laws. We would have loved to include prohibition of transfer of laws, as transfer in most cases end up in the wrong hands, and it needs to be prohibited. However, in the meantime, we are comfortable with the way the text is captured. Thank you. Chair [52:38]: Next on my list is the distinguished representative of Austria, to be followed by Israel and Colombia. The distinguished representative of Austria, you have the floor. Austria [52:53]: Thank you very much. And we will also just have a few comments on 1 to 5 and come back later on the details on currently, which is currently para 6. Starting out with just para 1, we noticed that there has been a change in the first sentence that we changed "employed" to "use". I think we would have a preference to "employed" since this is also the phrasing that we know from other protocols of the Convention, and it might be discussed what exactly is the difference, but for us it has a broader meaning and it could be more useful in this regard and would not give some misunderstandings that you could have weapons that you cannot use, but that you could employ or have kind of a different approach to them. The second point I wanted to make is that On, on the question of the formulation of these paras is that, uh, we, uh, note with, with appreciation that, um, with these paras, you have, um, looked into previous, uh, protocols of this convention and in many cases, uh, we are using the formulated formulation, it is prohibited, so this is quite a common formula, um, that we are using in the different protocols of the CCW. And I think it might be in line with our mandate that we are following these examples, but we are also open for other proposals. But for the moment, it's probably better to stick with what is on paper. Then there has been a question, Mark, I think, on para 3. Why is it just focusing on the civilian population and not on other aspects? And I think also here we have to look into the protocols that we are already having at hand, specifically Protocol III and Amended Protocol II, which has a very similar formulation. And the main idea, at least how we understand it, is that protection of the civilian population has such a meaning and also importance that this is something that high contracting parties wanted to reaffirm in a specific And this is also our reading of para 3. There are many other rules of IHL which are obviously protecting civilian populations, but I think this specific prohibition here, not making them the object of an attack by loss, is still valuable, and this is something that has been done with other protocols that we are having already. Our next point is that we-- I have noted, uh, your change when it comes to para five, which, which tries to take into account the discussion that we had about design, development, and, and, and use in that case as well, uh, I think it's a very elegant solution to, to, uh, solve our discussion on, in which paras we should also include the, the development phase. In this case, it takes into account the position of many countries that the main principles, especially those who are on interpretation of IHL, should be taken into account in the whole life cycle of the weapon systems. And this is also the notion that Article 36 brings to us. Article 36 does not does not take out the development phase. It's expanding the application of IHL into the development phase, and therefore, it is an important notion that we also have to include here, despite having it also in other parts of our text. As it has been already noted, there has been a deletion. We had a former para 6, which also looked into a prohibition of laws that operate without context, appropriate human control, and judgment. I think this is something we would still support, and we would like to-- have it back, um, then just maybe a quick remark without going into the detail, detail of para six, we also wanted to, um, ex- express that we would be strictly against, um, moving para six into box four. The main reason here is that what we are seeing in box three is the two-tier approach. It's a combination of prohibitions and regulations. It's the prohibition of certain autonomous weapon systems and the regulation of others. If we would just take the regulation part and put it into box four as national measures or something the states could do on a voluntary basis, we would basically destroy the tutti approach and take it out of the document. This is not in line with the working papers, the positions of many delegations in this room. I think it's important that those two elements are linked. And in that regard, but we'll come to that later, it's also important that para 6 is not a should paragraph, but that we should go back to need to or even must, so that it's very clear that those are regulations and not just some some interesting measures that states could undertake. This might seem strict to some delegations, but it also is important to point out that we have, as appropriate, inter alia and other disclaimers later on in the text. So we should have a strong start and then look into some softening of the language in A, B, C, D, E. So for the main part, it's important that this is a regulation para and not just a list of possible measures that states could undertake. I leave it at that. Thank you. Chair [58:43]: Let me thank the distinguished representative of Austria for sharing the position of his delegation. Next on my list is Israel, to be followed by Colombia and API-LS. But let me first pass the floor to the distinguished representative of Israel. Israel [59:02]: Thank you, Chair. Before we present our comments on box 3, we wish to further expand on our request to move paragraph 5 of box 2 to box 4, rather than placing it in box 3. With regard to the division between the boxes, it is our view that box 3 should focus on the fundamental legal prohibitions and restrictions in IHL applicable to laws, While box four should focus on measures to implement those prohibitions and restrictions. Therefore, in our view, box 3 is not the appropriate location for this paragraph, since human-machine interaction is not a primary rule of IHL, but rather a tool to implement IHL. Following the same logic, we have similar proposals regarding paragraph 5 and 6 in box 3, which we will present shortly. Moving on to box 3, regarding the third paragraph of the ruling text, We recommend preserving the original language of Article 51 of the First Additional Protocol that this paragraph draws from in order to ensure clarity regarding the legal obligation and to avoid further discussions on the implications of the deviations between this paragraph and Article 51's language. Therefore, the paragraph would read, "The civilian population as such, as well as individual civilians, shall not be the object of an attack by laws." Regarding the fourth paragraph in box 3, in our view, it is essential to stick to the language in Article 51 of the First Additional Protocol to the Geneva Conventions and to replace "anticipated and controlled" with the word "limited." Therefore, the paragraph would read, "It is prohibited to use laws if their effects in attack cannot be limited as is required by IHL in the circumstances of their use." The anticipation and control of the effects of an attack are not primary obligations under IHL, but rather second-order requirements intended to fulfill certain primary obligations, and the current box focuses on primary IHL obligations. We believe that the current text as it stands introduces a new legal standard, which is problematic. IHL requires to make a reasonable effort to try and assess collateral damage based on the information reasonably available prior to the attack, but does not require to predict it with certainty. Unpredictable collateral damage does not, as such, render an attack as illegal. The current wording does not reflect these important nuances and may be read as prohibiting many existing weapons or uses of weapons. The core point on anticipation of effects that does have a possible legal implication on the legality of weapons is sufficiently covered by the reference to the possibility of limiting the effects of the use of a weapon. In addition, predictability is addressed in other places in the rest of the text, such as in paragraph 6 in box 3 and in paragraph 3 in box 4. Chair, we're open to referring to anticipation of effects as a practical measure in the implementation of IHL obligations in a different part of the document, but not as it currently appears, namely, an independent legal requirement that by itself can disqualify weapons. This has no basis in IHL. Regarding the fifth paragraph in box 3, we suggest moving this paragraph to box 4, as it presents a practical measure, the consideration of IHL rules when developing a weapon, in order to maximize the chances that, when the weapon is used, it will comply with IHL. Regarding the sixth paragraph in box three of the rolling text, firstly, we would like to express our support for the Republic of Korea's suggestion to move paragraph six of box three to box four. This paragraph lists a variety of practical measures, some of which may be needed in order to implement IHL in certain circumstances, and some that reflect policy measures that exceed legal obligations. Considering the practical nature of this paragraph, the appropriate place for it is box 4, which outlines measures that support compliance with IHL rules. In line with the practical nature of this paragraph and with our previous comments regarding the term context-appropriate human judgment and control, we propose that the chapeau be phrased as follows, and I quote, "Measures, including during development, relating to control and human judgment are needed to ensure the use of laws is in compliance with the principles and requirements of distinction, proportionality, and precautions in attack, and other rules of international law applicable in armed conflicts. Such measures can include, where appropriate, the following, and so on. Regarding Article 6, while we can see a connection between predictability and reliability to IHL implementation, we're still not convinced that this is also the case with explainability and traceability. Therefore, we suggest deleting the words "explainable" and "traceable" from this paragraph. And regarding article 6, we find the coupling of assessments of legal and ethical considerations with the concept of reasonable chain of command and control to be confusing, and that it combines too many different concepts and terms. In order to streamline the text, we would propose deleting the second sentence so that it would read, "Ensure that laws are operated under a responsible chain of command and control." Regarding Article 6, as explained before, we objected the language "context-appropriate human judgment and control." Moreover, we believe this is even more relevant in this situation, as it is clear that there are alternative ways to ensure that the system does not make changes in mission parameters that could have negative legal implications without requiring real-time human control. For instance, the system can be designed to reprioritize among predefined categories of targets with a restriction on introducing new ones. Therefore, we suggest modifying the end of the sentence so that the text would read as follows. Ensure that laws, mission parameters with regard to types of targets, duration, geographical scope, and scale of the operation cannot be modified by the system in a way that will violate IHL. As for Article 6, we don't see sufficient justification for the current restriction. There is no need to differentiate between the ability to adequately identify a military objective by its nature and to adequately identify other military objectives, and we don't understand the basis for such a distinction. All military objectives must meet the condition of legality, and there is no justification for restricting the use of weapons against military objectives that meet these conditions. Therefore, we suggest the following wording, quote, limiting the use of laws against types of targets, for instance, by considering restricting the use of laws against objects that are not military objectives by nature, end quote. Thank you, Chair. I apologize for being a bit long, but we will share this in writing too. Chair [1:06:39]: I thank the distinguished representative of Israel, and indeed, the Chair would appreciate to receive the comments in writing. Next on my list is Colombia, to be followed by AP-ILS, Ireland, the United States, and Canada. But I will first pass the floor to the distinguished representative of Colombia. So you have the floor. Colombia [1:07:03]: Thank you, Chairperson. Colombia. aware that delegation of critical functions in lethal autonomous weapons systems poses significant risks of unpredictability, opacity in decision-making models, and vulnerability given unforeseen conditions, which jeopardizes the stability and accountability of the system. Which is why, for our delegation, it is indispensable that decisions on the use of lethal force be under human control and judgment, guaranteeing oversight, responsibility, and abiding by the principles of IHL. Our delegation supports the rolling text on paragraphs 1 and 4 of box 3 on the central provisions, which reflects imperative rules of international humanitarian law. We also support the initiative presented by the ICRC, under which it is recommended in paragraph 4 of this box to replace "controlled" with military given this reflects the existing terminology in IHL and it avoids ambiguity about what control means. Chairperson, regarding paragraph 5, we appreciate the reference to design, development and use. As several delegations highlighted during the March 2025 GGE meeting, extending coverage to development reflects the IHL requirements, including the requirement to respect IHL in all times and the obligation that the parties to an armed conflict be constantly mindful of protecting the civilian population, not just during attacks, but throughout the cycle of military operations. For our delegation, having greater clarity in the definition of the functional and non-functional requirements from the design and development stage of loss is indispensable and will allow us to take proactive ethical considerations that are founded on human judgment and control. On ethical considerations found in 6B, We appreciate the shared working document presented by Brazil in April 2025 on the use of the first global ethical standard for robotics and ethically guided automation systems developed by the IEEE as a technical reference framework to integrate ethical principles in the development and use of laws. Thank you. Chair [1:09:36]: I thank the distinguished representative of Colombia. Next on my list is the API-LS. You have the floor, sir. APILS [1:09:47]: Thank you, Chair. In our view, box 3 helpfully reflects a number of existing rules of IHL and then goes on to articulate practical measures that should be taken to ensure compliance with those rules. In light of this structure, the intent of paragraph 4 remains unclear. In our view, paragraph 4 conflates an existing legal obligation and the conditions or measures necessary to comply with that obligation. IHL prohibits the use of weapons the effects of which cannot be limited as required by law. And this prohibition is reflected in Article 51 of Additional Protocol I. Compliance with this rule appears to presume the ability to anticipate and to control the effects of the weapon. But this language goes beyond the rule itself. The concept of anticipation is not foreign to IHL, as the ICRC has pointed out. But existing IHL does not use that concept as part of the rule that we are discussing here. We suggest resolving this issue by reflecting the substance of paragraph 4 elsewhere. First of all, we note that paragraph 1 already reflects the underlying prohibition of indiscriminate weapons. A sentence could be added to the end of this paragraph, and I quote, "This includes weapons, the effects of which cannot be limited as required by IHL." End quote. Second, we note that paragraph 6 contains a series of measures for ensuring compliance with the law. So the language relating to anticipating and controlling effects would work better here. And while moving that language, it might also be prudent to replace the word "sufficiently" with the word "reasonably," because this is a term that is better understood in the IHL context. So we suggest adding a new subparagraph to paragraph six to say that states should, quote, ensure that they can reasonably anticipate and control the effects of laws in the circumstances of their use." With these changes, paragraph 4 would become redundant and could be deleted. Chair, we also suggest a minor amendment to paragraph 5. This paragraph currently refers to the prohibitions set forth in the preceding paragraphs. However, the preceding paragraphs do not only contain prohibitions. In particular, paragraph 2 contains a positive obligation to comply with certain IHL principles and requirements. Paragraph 5 could more accurately refer to the prohibitions and requirements set forth in the preceding paragraphs. With your permission, Chair, we would like to return to paragraph 6 later in the discussions. I thank you, Chair. Chair [1:13:00]: I thank the distinguished representative of IPI, IAPILS, sorry, and please send us your suggestions also in writing. Next on my list is Ireland, to be followed by the US and Canada, but let me first pass the floor to the distinguished representative of Ireland. So you have the floor. Ireland [1:13:21]: Thank you very much, Chair. This box is critical for putting into effect the two-tier approach of prohibition and regulation, which we welcome as the organising approach to these elements. Before I focus on the prohibitions in this contribution, in paragraphs 1 to 5, and reacting to some of the suggestions made in the room, I would like to support the comments made by Austria on retaining paragraph 6 in this box and changing the "should" in that paragraph to "must". This preserves the logic of the two-tier approach, which is extremely important for our delegation. As others have done, I will now focus on paragraphs 1 to 5 at this stage. Paragraphs 1 to 4 on prohibitions are well structured, reflect established IHL principles as they relate to laws, and build on previous understandings of the GGE, and we support their inclusion as prohibitions as they stand. On paragraph 5, we appreciate how the design and development phases have been taken into account. As we and others have argued, IHL consideration should be present throughout the system's life cycle, particularly due to the fact that consideration of core IHL principles must take place in the programming phases of development. We do not share the view that prohibiting design and development of such systems would hamper technological development more generally. Technological development can, should and must take place, but not in weapons systems that cannot comply with international law. In our view, the element we believe to be missing in these paragraphs is the reference to context-appropriate human judgment and control that was included in paragraph 5 of previous texts. We would prefer its inclusion, as others have suggested, and it would be consistent with our mandate to develop previous understandings in the GGE in light of the nature of these weapons. We're flexible as to where this could be reinserted or included in paragraph 1, 2, or as a standalone paragraph. Given that paragraph 4 of box 2 recognizes that context-appropriate human judgment and control is necessary to ensure the use and effects of laws are in compliance with IHL, then it logically follows that laws that operate without context-appropriate human judgment and control should be prohibited. While such a prohibition may be inferred by a combination of paragraphs 1 and 4 in this box, reinserting a direct prohibition would remove any ambiguity. Thank you. Chair [1:15:53]: I thank the distinguished representative of Ireland. Next on my list is the US, to be followed by Canada. But first, the distinguished representative of the US. You have the floor. United States of America [1:16:07]: Thank you, Mr. Chair. First, on box 3, my delegation wanted to recommend removing paragraph 6 and combining elements of it with the lists of measures in boxes 4 and 5. In our view, box 3 articulates the key legal rules that apply to the use of all lethal weapons, the principles and requirements of distinction, proportionality, and precautions in attack. We do not see paragraph six as articulating an existing rule of IHL or an appropriate new rule of IHL. Moreover, paragraph six has a number of problematic elements that we cannot accept. That said, we welcome the opportunity in box three to put the specific rules of IHL that will tell us what control, involvement, intervention, judgment, or related measures are required. As we have consistently indicated, we are open to building this terminology into these elements. For us, however, it is vitally important that we use these terms in a way that accurately orients them around legal principles and measures that can be understood and implemented by states. On paragraph six, we will also follow the lead of other delegations in coming back with more specific comments. So now I'll address paragraphs one through five of box three. Moving to paragraph two of box three, most IHL rules do not apply to the use of weapons. Another issue for us is that not all states are parties to the same IHL instruments. So we propose to delete the phrase IHL including from this paragraph. We can also add and other applicable IHL rules at the end of the sentence if that is helpful for other delegations. So the paragraph would read, laws may only be used in accordance with the principles and requirements of distinction, proportionality, and precautions in attack and other applicable IHL rules. Regarding paragraph 3, we first have a legal concern because it is not prohibited to make civilians in all circumstances the object of attack. In particular, civilians who take a direct part in hostilities forfeit their protection and may be made the object of attack. This issue could be fixed by adding "protected as such" after "individual civilians." And that was the solution that was used in the Arms Trade Treaty. Our second legal concern is with the phrase "object of attack by laws." Above, we are defining laws as weapon systems that select the target. Thus, one reading of this paragraph is that the laws will have an object of attack. This would be a radical change to the principle of distinction. For us, it is fundamental that the principle of distinction applies to persons and laws cannot have an object of attack. The person directing the attack makes persons or objects the object of attack. We've made proposals in the past to address this issue. Our proposal from March based on CCW Amended Protocol II, which we continue to support or can continue to support, reads, It is prohibited in all circumstances to direct attacks involving the use of laws against the civilian population as such, individual civilians protected as such, or civilian objects. In addition, trying to be more surgical, we can also support the following. It is prohibited in all circumstances to use laws with the intention to make the civilian population as such, individual civilians protected as such, or civilian objects the object of an attack. Regarding paragraph four, we see some promising ideas and want to work with the language here, but we don't see the existing language as a new rule of IHL. Rather, we see the language here as best an elaboration on the legal requirements in paragraph one, and in particular, on the concept of inherently indiscriminate weapons. So one recommendation we have is to move this paragraph right after paragraph one, because it is related to that one. You know, in addition, we think it's useful to say that inherently indiscriminate weapons include those that are designed to target civilians or civilian objects. So we think that's a useful prohibition that's reflected in existing IHL. We also think it is important to specify that this paragraph relates to the principles of distinction and proportionality rather than all of IHL. All of IHL does not apply to attacks. And lastly, we think it's important to change the circumstances of their use to any circumstances in accordance with the concept of inherently indiscriminate weapons. Therefore, we recommend that the paragraph read, inherently indiscriminate laws include those that are designed to target civilians or civilian objects and those whose effects in an attack cannot be anticipated and controlled as required by the principles of distinction and proportionality in any circumstances. Next, my delegation wanted to propose adding language on proportionality and precautions in attack. The GGE has previously reached consensus on the importance of these principles and requirements. but has not adopted specific language on the-- pardon me. But has not adopted specific language on the proportionality prohibition on excessive attacks or the obligation to take feasible precautions in attack. In our view, it would be useful to reaffirm these rules in the context of laws because they are among the core IHL requirements. We think these rules matter because they directly inform the measures that need to be taken and will be articulated in boxes four and five. In particular, we think it's useful to articulate the rule on precautions in attack because many of the measures we have been discussing can be framed in terms of precautions. We also think that the prohibition on attacks expected to cause excessive civilian harm is closely related to the concerns that delegations are raising about predictability and the anticipation of the effects of laws. I will not read out our proposals again, but again, we're happy to send these again to the Chair for consideration. Regarding paragraph five, we propose to begin this paragraph, "A State developing a laws needs to take into account rather than the current formulation, which is in the passive voice. This approach will be more consistent with some existing GGE language, which relates to measures taken by the state. We also propose to add in restrictions on the use of laws after prohibitions. This addition better reflects the legal rules above, which include restrictions as well as prohibitions. We also propose to move the notion of the use of laws in this paragraph. Our concern with the current formulation is that IHL rules already apply to use, and to say they may be taken into account into use seems to water down the existing legal requirements. Lastly, I'll note with regard to this paragraph that we see design as included in development, and so not necessary to specifically measure -- mentioned here. Therefore, the paragraph would read, "A state developing a laws needs to take into account the prohibitions and restrictions on the use of laws set forth in the preceding paragraphs," period. Thank you, Mr. Chair. Chair [1:25:32]: Let me thank the distinguished representative of the United States for making concrete proposals. Next on my list is Canada, to be followed by Panama, Japan, and Singapore. But let me first pass the floor to the distinguished representative of Canada. So you have the floor. Canada [1:25:47]: Thank you, Chair. My delegation would like to offer the following comments on box 3, paragraphs 1 through 5, and we'd like to make an additional intervention specifically on paragraph 6 at a later time. So in box 3, my delegation accepts the wordings in paragraphs 1 and 2 as presented. Moving to paragraph 3, my delegation would recommend removing the terminology "in all circumstances." This does not take into account the possibility of civilians losing their protective status by directly participating in hostilities. We could propose the following wording: "It is prohibited to make the civilian population, individual civilians, or civilian objects protected as such the object of an attack." On to paragraph 4, we accept the wording here as is. And finally, on paragraph five, my delegation wishes to underscore that the obligation to comply with IHL applies to the use of weapons, means, and method of warfare. It is our position that IHL compliance is required in the use of laws, i.e. complying with the principles of distinction, proportionality, and precautions. However, during the design and development stage, this is a matter for consideration as opposed to a specific legal requirement. I thank you. Chair [1:27:31]: I thank the distinguished representative of Canada. Next on my list is the distinguished representative... I would like to give the floor to the distinguished representative of Panama, thanking the distinguished representative of Canada. Madam, you have the floor. Panama [1:27:50]: Thank you very much, Chairperson. for giving me an opportunity to offer some additional comments on box 3. In paragraph 5, though we have said that we support the paragraph as drafted, we are able to support the proposal made by Malawi on transfer. On paragraph 6, the chapeau, we support the comments of Austria and Ireland to replace "should" with "must." And as we were saying earlier, the wording "context-appropriate human judgment and control" is as flexible as my delegation can be. which is why we prefer that that terminology be maintained throughout the text, with the understanding that it is a compromise solution proposed by the chairperson as a way of finding some common ground with those that have defended the notion of significant human control, among other concepts. Paragraph A, we would like to support it as drafted, especially to keep traceable and explainable in 6A. In 6B, we would like to support keeping the reference to ethical considerations, seeing as this is the only paragraph of the whole rolling text that has this reference. Additionally, it should be pointed out that this concept is found in the chapeau of the guiding principles. Further, given that autonomous weapons systems can select and engage targets regardless of human intervention, it raises serious questions about the role of humans in the use of force. These are all questions, all concerns that legal arguments in and of themselves cannot answer. So it's not a matter of getting into a race of the legal body, of the legal context. Rather, it should offer a source of legal discussion of the existing content and the content of the regulation, which is why we would like to keep that reference in the text. We would also like to keep paragraph 6 and its subparagraphs in their entirety, given that it complements paragraph 4 of box 3. Those are our comments for now. Thank you. Chair [1:30:51]: I thank the distinguished representative of Panama and let me now pass the floor to the distinguished representative of Japan to be followed by Singapore. Japan [1:31:03]: We would also like to comment first on paragraph 1 through 5 and come back to paragraph 6 later. I have two points. On the development, we share the necessity to consider the whole life cycle of a weapon system, including development, as well as the importance of predictability and reliability of effective use of a weapon system. At the same time, we should be mindful of not creating excessive chilling effects to development for peaceful purposes, as well as development for security purposes, that does not cause any problem with regards to ensuring compliance with international law. From this perspective, on paragraph 5, while we agree with the content and the language, we would agree with India to move it to section 4 as it relates to considerations in the development phase and others. Also on this point, we support the US's comment to delete use from this sentence as use is addressed in other prohibitions. Secondly, on paragraph 4, we support the proposal by Israel to replace "anticipated and controlled" to "limited," which, in our view, would align better with the IHL, especially the Additional Protocol of Geneva Convention, Article 51. Thank you, Mr. Chair. Chair [1:32:46]: I thank the distinguished representative of Japan for her comments. Next on my list is the distinguished representative of Singapore. You have the floor. Singapore [1:32:59]: Thank you, Chair. Regarding the proposed measures under discussion in box 3, we currently have the following comments. On box 3, paragraph 2, my delegation is supportive of the revised paragraph, which helps to clarify the IHR obligations on states. That said, to make clear that these are obligations, we propose that "may only" be replaced by "must". The text would thus read, "Laws must be used in accordance with IHL, including the principles and requirement of distinction, proportionality and precautions in attack." On box 3, paragraph 6, my delegation would like to raise two points. First, During the June informal consultations as well as the March session of the GGE, my delegation had posed questions about the requirement for an assessment of legal obligations and ethical considerations by a human. My delegation is of the view that legal obligations are defined by an established and codified body of law agreed upon by States Party, for example, the Geneva Conventions of 1949 or the 1977 Protocols to the Geneva Conventions. However, ethical considerations are normative considerations of what is right and what is wrong, for which there is no agreed set of moral standards among States. For avoidance of doubt, this is not to say that ethical considerations are not worthy of discussion. Nonetheless, we would like to seek clarification on how to make ethical considerations a military requirement in practice, when such normative considerations are not agreed on between States. Second, paragraph 6 further specifies that these assessments should take place in particular with regard to the effects of the selection and engagement functions. My delegation's view is that this specification may not be necessary, as the focus of the legal assessment should be on States' compliance with IHL obligations. Given the above reasons, My delegation is of the view that the second sentence on legal assessments and ethical considerations is not necessary, and it would suffice for us to retain the first line in box 3.6 to ensure that laws are operated under a responsible chain of command and control. That is all for now. My delegation will seek the floor again at a later stage to further nibble at the paragraph 6.3. I thank you, Chair. Chair [1:35:47]: I thank the distinguished representative of Singapore for sharing the views of his delegation. Next on my list is the distinguished representative of Mexico, to be followed by Switzerland. But first, madam, you have the floor. Mexico [1:36:05]: Thank you, Chair. At the outset, I would like to once again Thank you for your work. Thank your team to prepare this compilation and, in other words, summary of the discussions that were held by the group and in this box and in this list of prohibitions. For my delegation, this is actually the very heart of what may become a future instrument under the CCW. So it is very important to keep these prohibitions in a clear, direct manner, the strongest possible way. And we listened to the recommendations that were made by other delegations. for example, in six to change the word should and replace it with must to strengthen the commitments of states and also not to change the prohibitions that already are contained in connection in the text in connection with IHL and the need to respect the principles of distinction, precaution and proportionality always. And finally, we also believe that these prohibitions are quite direct, and we believe that this will strengthen the text and this will help us resolve any doubts whether a law system can function within the framework of IHL. As we've seen from the document of the ISARC, this is something that is more and more difficult to do, but it is indeed necessary for us to define how we do that in these discussions and eventually during the negotiations. Thank you, Chair. Chair [1:38:30]: I thank the distinguished representative of Mexico. Next on my list is Switzerland, to be followed by Argentina. But let me first pass the floor to the distinguished representative of Switzerland. So we have... Switzerland [1:38:45]: Thank you, Mr. Chair, for giving us the floor. I would like to make a few comments on paragraphs 1 to 5, and we would come back to paragraph 6 at a later stage. As a way of introduction, I would say we can support paragraphs 1 to 5 as they stand. Perhaps a little bit in reaction to what have been said by different delegations, we would be having certain reluctance to change language and rules that have been accepted previously, for instance, with regard to the prohibition of direct attacks against the civilian population as such, individual civilians or civilian objects, where formulations exactly in this sense have been accepted in different protocols of the CCW, for instance, protocol two or amended protocol two or Protocol 3, all of which contain this formulation. On paragraph 2, it would seem important to us to not limit the statement only to the fundamental principles of distinction, proportionality and precaution, as a wide range of rules will be relevant to the use of lethal autonomous weapon systems for the conduct of hostilities, but also beyond. With regard to paragraph 4, we support the reference to anticipation and controlling or, as some have said, limiting the effects. Those would be important characteristics for us. They show what is needed to comply with IHL. For instance, predictability and anticipation of effects are in our understanding a necessity for being able to comply with IHL and to make the relevant assessments. For instance, the prohibition of indiscriminate attacks includes those which employ a means or methods of combat the effects of which cannot be limited as required by IHL, or the principle of proportionality explicitly balances expected incidental harm with the concrete and direct military advantage anticipated. And finally, on paragraph five, we can support some feelings that has been expressed in the room that the paragraph should not be limited only to corporations, but also and be added to terms like requirements or similar, as not all points above solely refer to prohibitions. Thank you, Mr. Chair. Chair [1:41:47]: Let me thank the distinguished representative of Switzerland. Next on my list is Argentina, to be followed by Sweden, Australia, Italy, and Kazakhstan. But let me first pass the floor to distinguished representative of Argentina. So you have the floor. Argentina [1:42:06]: Thank you very much, Chair. Since it is the first time I'm taking the floor, I'd like to once again thank you on behalf of my delegation for your efforts to prepare the update of the rolling text. We believe that we have moved quite far along the path as we work on this text and could be a good foundation taking into account the mandate. And we will be ready to support this as we move forward and we will continue to be flexible in considering the text to achieve that goal. As far as boxes one and two are concerned, our quite acceptable to us and represent a very good balance between the different positions. Now, on box three, we believe that this is a very positive approach, trying to avoid any conflictive elements. We see that in the beginning, we have the basic prohibitions related to these weapon systems, and we participated in the work of the Group for preparing a possible Protocol 6 on these questions. Now, I would like to make a comment on paragraph 6 and small Roman 4. This was language that was added to the version that we discussed in March, and we already expressed our reservations at that point. We believe that "military objectives" is already something that is referred to in the Geneva Conventions. We believe that still this term is not enjoying full consensus because we have to see that objectives may represent an absolute category that can be subject to different interpretations. But once again, we would like to reiterate that we remain flexible in our position. Thank you. Chair [1:44:33]: I thank the distinguished representative of Argentina. Let me now pass the floor to the distinguished representative of Sweden. You have the floor. Sweden [1:44:43]: Thank you very much, Chair, and good afternoon, colleagues. Sweden appreciates and support the changes made in this box based on our discussion in March, and we believe that this is moving us in a good direction. Like many other delegations, we support a two-tier approach with prohibitions and restrictions and want to keep that structure in the rolling text. And as long as the rolling text reflect this structure, we can be flexible whether, for example, para 6 should be in box 2 or box 4. We appreciate that paras 1 to 4 place existing IHL prohibition and requirements in a clear laws context. Para 3 is important, but as the U.S. delegation stated, we wish to note that civilians are protected unless and for such time as they take direct part in hostilities, and we therefore support the U.S. suggested addition of "protected as such." More generally, we are open to some modifications relating to language, but prefer treaty near formulations regarding para 1 to 4. For example, para 4 could be formulated along the lines of Article 51, para 4 of Additional Protocol I to the Geneva Convention. We appreciate that para 6 sets out measures that exemplifies on how humans may exercise judgment and control to ensure that laws can be used in line with international law, in particular IHL. And finally, to us, it is important that this para maintain, inter alia, or the similar wording thereof. I thank you, Chair. Chair [1:46:22]: I thank the distinguished representative of Sweden. Let me now pass the floor to the distinguished representative of Australia. Madam, you have the floor. Australia [1:46:32]: Thank you, Chair, for giving me the floor. We consider that paragraphs 1 to 4 of box 3 are expressions of IHL, and we support their inclusion and retention. We are encouraged by the edits made to the rolling text during the intercessional period, which have increased the clarity and fidelity of these paragraphs to IHL. Chair, we echo the comments of Canada and prefer that "in all circumstances" is removed from paragraph 3 so that this paragraph accurately and clearly reflects IHL. On paragraph 4, we are interested in the suggestion of APILS to merge this paragraph with paragraph 6A. We see this as a possible landing zone. In respect of paragraph 5, we see the edits that have been made to this paragraph as a significant improvement on the prior version. We welcome the removal of the prior prohibition on employment of laws which operate without context-appropriate human control and judgment. As previously expressed, our view was that this concept was not an appropriate basis for a broad prohibition. We can support the current paragraph five as we agree that those aspects of IHL listed in paragraphs one to four of box three should be taken into account in the design, development, and use of laws. Chair, turning now to paragraph six, we support the edits to the chapeau as they clarify that context appropriate human control and judgment is a means to support compliance with IHL in the context of laws rather than a separate obligation of IHL. Australia supports the way paragraph 6 is framed, which seeks to enumerate specific measures for laws and give meaning to the concept of context-appropriate human control and judgement. We see these as measures specific to laws, akin to those set out in box 4. We will intervene separately on our understanding of context-appropriate human control and judgement, Chair, and the measures listed in paragraph 6A to D. I thank you. Chair [1:48:33]: I thank the distinguished representative of Australia. Next on my list is Italy, to be followed by Kazakhstan, the United Arab Emirates, and France. But now first, the distinguished representative of Italy, you have the floor. Italy [1:48:50]: Mr. Chair, thank you for your work on another relevant section of this rolling text, which includes some principles that have already been recognized by the GGE in the guiding principles and reiterate the application of IHL rules and principles to laws. We support the current wording, which distinguishes prohibitions from obligations and general prohibitions. We have some comments on the following paras. In para 4, in terms of "controlled", as other delegations already pointed out, we would prefer to replace it with "limited", in line with Article 51, paragraph 4 of Additional Protocol I. Furthermore, we would suggest that we add after "use" except unforeseen and unforeseeable effects related to exogenous causes that represent cases of force majeure. Anticipating the effects is not, in fact, always possible, especially those unexpected and unpredictable linked to causes outside of human control. And for this reason, we would recommend this addition. In para 5, we are glad for the addition of the terms "design" and "development" in line with Article 36 of the first additional protocol to the Geneva Convention. We, in fact, believe in the importance to preserve and retain context-appropriate human judgment and control to reconnect with the section 2 at all times and across the entire life cycle of the weapon. It is an aspect that we also had the opportunity to dwell on in the working paper dedicated to the human element we submitted with 10 other countries to this GGE session, and that the distinguished delegation of Germany detailed yesterday. In para 6A, we confirm our support to this sentence, especially in the provision of the facts of these weapons systems. The use of predictable, reliable, traceable, and explainable clarifies the actions taken into consideration for full compliance with IHL prior to employment. In para 6, we support the new wording focused on the chain of command. which is responsible for deciding, based on a legal or ethical assessment, the legitimacy of the target selection and engagement. This formulation reflects the criteria established in the selection process and therefore would appear to be largely acceptable to many delegations. In para 6C, We would suggest to add the word goals prior to the types of targets, rephrasing with the following formula. Ensure that those mission parameters with regard to the goals or types of targets, condition, duration, geographical scope, and scale of the operation cannot be modified by the system without context appropriate human judgment and control. In order to increase clarity, we suggest that we add a new sentence, "Laws cannot modify preset parameters without human intervention." This amendment has the purpose to limit the machine learning or deep learning capacity, resulting in analysis developed independently by the autonomous system. Regarding para 6, we would recommend eliminating this para pertaining to the limitation of the number of engagements, as this restriction appears to us unclear from a military point of view. Also, a numerical limit on the targets of a single autonomous system could determine on an operational level the need to have multiple systems in order to engage a greater number of targets, with the risk of encouraging the proliferation of lows. Finally, on para 6, we would like to recommend a language more in line with IHL provisions, Rephrasing the first sentence in, uh, restrict the use of laws against objects that are military objectives by nature or otherwise lawful military targets, limiting the use of laws against military targets may be unclear. ICHL describes in the Additional Protocol I, Article 52, para 2, different types of objects that contribute to the military action and constitute legitimate targets by their nature, location, purpose, or use. And of course, Chair, we are going to send our recommendations into writing. Thank you. Chair [1:54:24]: Let me thank the distinguished representative of Italy and also for the offer to send the remarks in writing. Next on my list is Kazakhstan, and then we have the United Arab Emirates and France. But let me first pass the floor to the distinguished representative of Kazakhstan. So you have the floor. Kazakhstan [1:54:41]: Thank you, Mr. Chairperson. On box 3, our delegation would like to highlight several issues. several parts to be commented. The part of five, we propose to keep or even enhance this part and obviously to retain wording about design, development and use of laws because all these elements should be taken into account during the implementation of such weapon and the prohibition that's set in preceding paragraphs. In chapeau of para 6, we propose to beg for the words of meaningful human control because this context appropriate human judgment to control it. not so clear and can create, let's say, ambiguous connotation of what is it context, what is it appropriate. So that they're using meaningful human control. It is a long-standing definition that used to be for a long time. And the last comment is about 6B and on retaining and on ethical consideration. Moreover, we prefer to use ethical and ethics reference in other boxes as well, but we are talking about this. So our preference to keep this ethical consideration words in the whole 6B. And as a part of interactive discussion, so we have to be assured that legal obligation will be taken into account. But the ethical is more than legal, more than IHL. In fact, ethical issues is a basis for all humanitarian conventions, all humanitarian documents. And the IHL based on the, first of all, on ethical considerations. Ethics is necessary to guide, to limit, to humanize the such functions of this kind of weapon. Thank you. Chair [1:57:48]: Let me thank the distinguished representative of Kazakhstan. Next on my list is the United Arab Emirates. Let me pass the floor to the distinguished representative of the United Arab Emirates. United Arab Emirates [1:58:05]: Thank you, Mr. Chairperson, for allowing us to take the floor once again. I would like, in my intervention, to present observations regarding paragraphs 1 to 5, and I shall tackle other paragraphs at a later stage. Regarding paragraph 1, we concur with the general principle of prohibition in the text, and we propose to to mention in a clear way the Additional Protocol of the Geneva Conventions. Regarding paragraph 2, it is a repetition of paragraph 1, and we propose that it should be merged with paragraph 1 so that we have one paragraph without duplication or repetition, and it will also send a clear message on the principles of distinction, proportionality, and precautions. one and two tackle the commitment to the principles of IHL, namely principles of distinction, proportionality, and emerging of both paragraphs will bring a more clarity on principles, procedure without any repetition. And this is what is recommended when it comes to legal text. Regarding paragraph 3, we concur with this principle, which is the foundation of IHL, and there is no need for any amendment. And we recommend to keep this paragraph as is. And we also reiterate the importance of this drafting that calls for prohibition of indiscriminate attacks due to attacks by laws. Regarding paragraph 4, we concur with this direction because it continues in the same track of human control. and we prefer to replace "anticipated" and "controlled" by a clearer drafting. Regarding paragraph 5, we concur with the content of this paragraph, and we have some reservation on "context appropriate", and we prefer to refer to human judgment and control, and we recommend to merge paragraphs 4 and 5 so that we can bring in more clarity about human control and commitment to IHL. And we also think that this merging of the two paragraph will mean that absence of control will mean absence of judgment. However, I would like to reiterate that paragraph 5 is very important when it comes to compliance with IHL. The legal assessment at the level of design, development, and use is in line with the first additional protocol of the Geneva Conventions, where countries are obliged to have a legal assessment when it comes to international obligations. That's why we would like to keep this paragraph with a clearer drafting so that it can include a clearer formulation when it comes to distinction, proportionality and precautions, as well as assessing the levels of human control and securing that technological developments will not lead to unpredictable developments. This is in line with the positions of many countries that called for early insertion of legal requirements when it comes to the life cycle of laws. And this reflects a common commitment to prevent any violations prior to any attack. Thank you. Chair [2:02:41]: Of the United Arab Emirates. And next on my list is France, to be followed by Hungary. But let me first pass the floor to the distinguished representative of France. you have the floor. France ยท Ambassador [2:02:55]: Thank you, Chair. I would like to offer some comments on paragraphs 1 through 6C. We support, as are, paragraphs 1, 2, 3, and 6A. In paragraph 4, we acknowledge what it does with certain concepts in the first Additional Protocol and how it approaches that. So we would be willing to replace control with limited, as proposed by the ICRC. In paragraph 5, it is our view that it would be preferable to replace prohibitions with principles and requirements of IHL. given that paragraph 2 does not deal with a prohibition. Paragraph 6, we can support replacing should with need to, given that it is our belief that human control arises from certain obligations of IHL. We also understand certain delegations' desire to move this paragraph to box 4, given that, actually, as drafted, it seems to be devoted to certain measures to be implemented. In 6, like comments made by Singapore, we call for the removal of ethical considerations given that IHL only requires to take into account legal criteria and operational considerations. In paragraph 6, though certain amendments or machine learning can be part of autonomy, the parameters of the mission must not, however, be modified significantly. outside of the scope of use defined by the command chain, which is why we agree with this wording, which refers more to that use case, which is why we believe that it is essential to add significantly before modified. Given that autonomy assumes the capacity to adapt to the environment, If I may, later on, I would like to take the floor again to speak about 6 and its subparagraphs. Chairperson, thank you. Chair [2:05:51]: Let me thank the Ambassador of France for sharing the views of her delegation. And of course, you can always come back later in the discussion, as many have already announced that they will do with regard to paragraph 6. Let me now pass the floor to the distinguished representative of Hungary, who will be followed by Austria, but first Hungary. Hungary [2:06:11]: Mr. Chair, since it is my first time to take the floor, my delegation wishes to congratulate you and your team on guiding the work of this group of governmental experts. My delegation would like to propose in box 3, paragraph 3, the following. It is prohibited in all circumstances to make the civilian population as such individual civilians or civilian objects, and following our proposal, other individuals, for example, medical personnel, prisoners of war, and objects protected by international law and international humanitarian law, as an example, field hospitals, cultural heritage, installations containing dangerous forces, in accordance with Article 38 and 56 of the Additional Protocol of the Geneva-- Protocol of-- First Protocol of the Geneva Convention, the object of an attack by laws. Thank you, Mr. Chair. Chair [2:07:10]: I thank the distinguished representative of Hungary. Next on my list is Austria. So you have the floor. Austria [2:07:18]: Before going into the second part of our intervention on this box with regard to para 6, we also would like to come back to some of the other paras that have been currently discussed. Just going back to para 2, I think this is extremely important, especially that we are referencing specifically these principles. Many of these key paras that we are setting out here, they are reiterating key principles of IHL. Of course, not all principles of IHL are reiterated here. We are focusing on those who we, as high contracting parties of the CCW, think have a very specific importance when it comes to these kind of weapon systems. I think this is the main idea here and these principles that are set out here in two are specifically important and they are, of course, also part of customary IHL. I also have to say, unfortunately, that several proposals that were here for the United States are going lower than obligations that many states have under Additional Protocol I. which is also considered customary IHL by many. And it would not work for many states to have weaker rules in a specific instrument than the general rules that they already have to comply with. So this is a concern that we are having with some of the proposals that we hear. Another issue that has now come up in the discussion is, I think, different readings of para 3. And if we look into the text specifically and what it includes and what it does not include, I think the problem is that the previous drafting was taking example from other protocols, specifically Protocol, Amended Protocol II and Protocol III, where we have this specific protection by the civilian populations. In these two protocols, As far as we interpreted, the idea was to give civilians a special status of protection that maybe goes beyond the rules of IHL that apply in general, and that might create this confusion here. So the question that we have to answer to ourselves is, do we just want to reiterate the rules that apply to all kind of weapon systems, or do we want to go further, as we did in AP II and Protocol III, and have a stronger and also wider reaching protections of civilians that applies to them at all times, even if they are forfeiting their protection rights. This is a question to be answered, and it's probably not good that we have different interpretations what this article means, but we should look deeper into it. I think we are fine with the formulation as it is right now, as we have been also in our previous discussion in the last session, but it's definitely something that we should have clarity what in the end, what does it do in the end. On para 4, We still think that anticipated and controlled have their usefulness, as the ICRC has set out. This is something that can be found in IHL, and we are also of the opinion that this is something that we can include for the purpose of prohibiting or regulating laws, as it is a specificity of these weapon systems that it's not always clear that you can anticipate and control them. With conventional weapons, there might be some mistakes or some colleagues also referred to a force majeure or something like this. But the problem is that there is an inherent risk with algorithmic systems that they could not behave in an anticipated way. And this is one of the risks that we want to address in this para by also referencing existing IHL I mean, the limitation argument is also valid. Maybe we could further include it here, because this also relates, of course, to Article 51. On para 5, I think we already discussed this in our previous intervention. We just wanted to go to underline that this is to a larger than an existing obligation under Article 36. So what we are doing here, we are reiterating it and specifying it in more detail. But what we also have to point out that Article 36 does not only refer to the principles and requirements under IHL. It also refers to any other rule of international law applicable to the high contracting party. So this also means international law in general, but it would also apply to the other prohibitions in a possible protocol. So we should not go below that. And if this is an interpretation or a reiteration of Article 36, we cannot have a lower standard as it is set out there already, and I think this applies to all the parts in this box in general. On para 6, we have already expressed our opinion that this should stay here. If people -- and this is maybe similar to the discussion that we had on box 2. If people feel that this should be somehow separated from the principles at the beginning of this box, we can also make a new box on para 6, so we have another separate box on all the elements that come in para 6 and beyond. I don't think that it's necessary to mix them with box 4 because this is just a different approach than all these measures that we have laid out in box 4. It is the regulation part, and it is an important and essential element of the two-tier approach. On 6A, we would like to keep this para. It's quite important, and we had a lot of discussions on all of these elements and what kind of purpose they serve. And we are here talking about a measure where we talk about ensure. on 6B, we would like to keep the ethical considerations. I think they're extremely important. It is a pity that this is the only part of the text where this is actually used. This is a big flaw of the current approach that we are taking, and it should definitely be maintained here. We are also interested to keep the effects of the selection and engagement functions because these are the main functions where an ethical assessment is an ethical consideration and also the general assessment is the most important one. And I think people don't want to have unethical considerations by a human on other functions of an autonomous weapon systems. I think these are the most important one. On 6C, there was a proposal to delete without context appropriate human judgment and control. I think this would unduly broaden the scope of 6C. But to some extent, it's also possible. I mean, what would broaden the scope was the proposal that we have an alternate formulation that refers to IHL in general. What we could do is we could stop the para just after system and have a very general formulation, not referencing the context-appropriate human judgment and control, and just make very clear that mission parameters should not be changed. On D, we would like to keep Roman II. This is extremely important. We had a discussion about this. And I mean, we do not have to reiterate all the arguments, but in general, this is just something that more relates to the character of algorithms, which can have a lot of-- can do a lot of calculations, reiterations of the same function. in a small amount of time. So limiting the number of engagements is actually a way for humans to keep control over the system and not have 10,000 calculations in a few seconds. And I think this is in line with all the interests that military operators have in general. I mean, this is definitely a best practice that should be included in any kind of text that we are using here. We would also like to suggest to reintroduce a new Roman IV in front of the current Roman IV with limiting the types of targets. This was an element that was previously in the previous Roman IV. It was the previous formulation had restricted use of laws against objects that are military objects. by nature and limiting the types of targets in one Roman IV paragraph. In the new version, the limiting of types of targets was deleted, but we believe that this could be its own Roman numerical. And I think it benefits from the separation of these two ideas. And it's just something that you can do before you put your autonomous weapon systems into operation. And then finally, we have a remark on Five, about real-time machine learning. I think we have different understandings here in this room what real-time machine learning entails or what it means. One understanding that we feel very close to is that this limitation applies to a live learning during the operation of the systems. And, uh, the possibility that the algorithm or the system can modify itself during the operation, operation and, and change, um, the, the way of working and the way of, uh, uh, um, the way of, of operation, um, this could pose severe risks. I think a misunderstanding that we have in this room is that what we are seeing here should not prohibit machine learning per se or during operation. I think the question is when does the real, when is the real time data fueled into the system? There has been some discussions and also some papers on this issue. Real time machine learning would not mean that that the system can change itself. So it's quite similar to what we have with changing the mission parameters, but here the question is about the system changing its functioning in general during the operation. This does not mean that real-time data cannot be channeled into the system, but it should be not during a live operation while being in combat and can be done later. The data is collected and it is channeled into the system at a later stage, and I think this would be a much more safe way of conducting this. If you're looking into this question in this way, limiting is definitely too weak, and we could even talk about prohibiting real-time machine learning. But of course, we have to come to a common understanding what real-time machine learning actually means, maybe also look into different formulations that we don't have misunderstandings what we mean with that word. Sorry for being so long, and thank you. Chair [2:19:16]: I thank the distinguished representative of Austria. I have three more requests for the floor, Algeria, Denmark, Spain. I think by now we have, yes, we take one more now. I think by now we have entered the discussion on paragraph six, so those who reserved their position Uh, to that, uh, please come now also, take the floor, show your plaque so we have a, a kind of, um, impression how many, uh, requests there will still be for the floor this afternoon. But I will now pass the floor first to the distinguished representative of Algeria. Madam, you have the floor. Algeria [2:20:01]: Thank you, Mr. Chair. My delegation wishes to deliver the following observations on the third box of the rolling text. We fully support the indication in the rolling text to the absolute prohibition of the laws that are inherently unlawful, those that cause unnecessary suffering, are indiscriminate by nature, or incapable of being used in compliance with IHL. In line with the fundamental principles of international humanitarian law, such as Article 13 of Additional Protocol II to the Geneva Convention, which established that civilian population as such, as well as individual civilians, shall not be object of attack, we wish to emphasize the critical importance of paragraph 3. Safeguarding civilians must remain the primary objectives under all circumstances. We agree with the core message of para 4 and we emphasize that the fully autonomous laws cannot meet the requirement of the anticipated effect and control and a machine will never possess the human judgment. We take note on the measures presented in para 6 of the rolling text and we believe that human control over machines throughout their entire life cycle is a key principle that should be respected. In addition, we see merit in ensuring that laws operate under a responsible chain of command and can be predictably controlled, including through restriction on the scope and duration of their missions and the ability for human operator to deactivate the system in timely manner. We caution, however, against reliance on autonomous self-deactivation mechanisms. as such measures would shift responsibility from the human operator to the machine itself. Effective human judgment and accountability must remain at the core of any use of these systems. Regarding paragraph 6C, we already expressed our position regarding the term context-appropriate human control and judgment. In line of this, we think that this term needs further explanation in compliance with IHL principles. I thank you. Chair [2:22:36]: I thank the distinguished representative of Algeria. Next on my list is Denmark, to be followed by Spain, ENCO and Singapore. But first, the distinguished representative of Denmark. You have the floor. Denmark [2:22:50]: Thank you, Chair, for allowing me to take the floor. Although we have now opened the discussion on para 6, we would like to limit this intervention to comment on para 1 to 5. On para 6, we would consider making a more detailed intervention at a later stage. But we can support paras 1 to 5 overall. Like a number of delegations, however, we would like to suggest additional language to para 3. So after individual civilians, we would add, unless and for such a time, they are directly participating in hostilities. The full sentence would thus read, "It is prohibited in all circumstances to make the civilian population as such, individual civilians, unless and for such a time they are directly participating in hostilities, or civilian objects, the object of an attack by loss." However, we could also accept language proposed by colleagues from Spain, Canada, or the United States to reflect this principle. And finally, we would like to echo the comments made by the distinguished delegate of Sweden in that we would like to see the two-tier approach reflected in the text, and as long as this is respected, we can be flexible on the placement of certain paras, including para 6. Thank you. Chair [2:24:10]: I thank the distinguished representative of Denmark for sharing her views and making proposals. And next on my list is the distinguished representative of Spain. You have the floor. Spain [2:24:27]: Chairperson, thank you for giving my delegation the floor once again in order to share our comments on paragraph 6. We support the wording of subparagraphs,, and, which is why we will be focusing our comments on subparagraph. On, we believe that it is essential to keep this paragraph as worded, given that it bolsters the position held by my delegation in favor of human control. Here, once again, I would like to refer to working document four on the human element presented by several delegations, among them my own. For Roman numeral two, we would like to propose that it be deleted. The reasoning for this is that we do not understand the need to limit the number of engagements of these armed systems if these engagements meet with international humanitarian law. On Roman numeral three, We believe that it is good wording. On Roman numeral 4, we would like to propose that "by nature" be removed. The reasoning for this proposal is that the concept of military objective in Article 52 of the Additional Protocol I includes not just those objectives that by their nature are military, but also objectives that because of their purpose, location or use contribute to military action. I would like to note that this comment is related to the proposal that we have made previously regarding paragraph 3. Lastly, as regards paragraph Roman numeral 5, we propose removing "target selection" in the current draft. in order to avoid hindering technological development, especially if said technological development guarantees full compliance with international law, and in particular, IHL. Thank you. Chair [2:26:43]: I thank the distinguished representative of Spain for commenting on the text. Next on my list is ENCODE, to be followed by Singapore. But first, the distinguished representative of ENCODE. You have the floor. ENCODE [2:26:58]: Thank you, Chair. As it's the first time we are taking the floor as ENCODE, we would first like to thank you and your team for your continuous effort on this running text. And then we would like to comment on paragraph six sub C and also refer immediately to the white paper on the human element written by a range of states and presented by the German delegation yesterday. From a more technical perspective, we would like to point out that human choices on the parameters of a mission have different effects in models that rely on supervised learning than in systems that rely on unsupervised or reinforcement learning. In models that are trained in a supervised way, humans label the data themselves in order to explain to the model which data points belong to which label. Unsupervised or reinforcement learning means that the data is fed to the system unlabeled, after which the model is given the assignment to find underlying or latent patterns that inform subsequent decision-making. Although humans point towards the areas in which patterns should be found, there is a much greater risk for the model to so-called hack the system. These risks include goal misalignment and reward hacking. Late last year, we have published a short paper including all the different ways in which a model can get a bit too creative with the parameters given to it. We would be happy to send that to the delegation who have delegations who haven't received it yet. Referring to the understanding of context appropriate human judgment and control in the white paper on the human element, this might create a loophole we want to cover. In the White Paper, it is stated that autonomy is inherently bound by human design choices. This creates the possibility for a developer of an unsupervised model that goes into the weapon system to argue that they made relevant design choices just by pointing at the areas in which patterns should be found. However, the autonomy of models based on unsupervised or reinforcement learning can hardly be adjusted to different contexts, as most of them are what we call black box algorithms. of which the explainability is rather limited and thus the control is rather limited. We propose to cover this by having the list of factors that comprise human judgments included in the White Paper to also include the risk of model inaccuracy and to acknowledge that autonomy is not inherently bound by human design choices but human design choices, including the choice on the type of learning, must be made such that autonomy is bound in a way that requires humans to judge whether a system can be used in compliance with IHL. We acknowledge that this would not lead to a change in the wording of the rolling text immediately. However, we do think that this is a risk that should be considered relevant when coming to a shared understanding of what entails context-appropriate human control and judgment in different places in the text. Thank you. Chair [2:29:53]: Let me thank the distinguished representative of Angola also for commenting on a working paper by a group of states. Next on my list is Singapore, to be followed by Japan and Canada. But now first, the distinguished representative of Singapore, you have the floor. Singapore [2:30:11]: Thank you, Chair, for giving us the floor again. My delegation continues with further comments on paragraph 6. On box 3, paragraph 6, d, roman numeral 4, similar to the distinguished delegation from Spain, my delegation had previously, during the informal consultations, queried about the proposed restrictions that laws may only be used against objects that are military objectives by nature. Under existing IHL, pursuant to Article 52, clause 2 of AP1, objects are deemed to be military objectives not only by their nature, but also by way of their location, purpose, and/or use where they make an effective contribution to military action, and whose total or partial destruction, capture, or neutralization in the circumstances ruling at the time offers a definite military advantage. Now, let us illustrate all that with an example. If a building compound, which was originally a textile factory, has been converted by the belligerent into a missile storage facility, The use of the civilian object for military purposes makes it a legitimate military objective under IHL. Similarly, the location of an object may affect its status as a legitimate military objective, regardless of its nature. For example, a bridge could be used by the belligerent for troop movement to reinforce frontal positions. The bridge, as a dual-use object, by virtue of its location, becomes a legitimate military objective under normal IHL rules. Given the examples above of realistic situations where objects might become legitimate military objectives under IHL beyond their nature, my delegation is interested to understand why this paragraph is scoped selectively to restrict the use of laws to military objectives by nature. Next, on paragraph 6, limiting real-time machine learning with regard to target selection and engagement functions, my delegation is of the view that the proposed articulation currently introduces a few ambiguous terms that would make implementation impractical. For instance, what would be considered real-time? What would the scope of machine learning constitute? Assuming the aim of paragraph 6 is to limit the ability of laws in autonomously altering the types of targets and parameters of engagement, then my delegation puts it to the floor that the current phrasing of paragraph 6 may already fulfil the intention of those limits. I thank you, Chair. Chair [2:33:09]: I thank the distinguished representative of Singapore. Next on my list is Japan, to be followed by Canada and Ireland. Let me now pass the floor to the distinguished representative of Japan. You have the floor, madam. Japan [2:33:27]: Thank you, Mr. Chair, for giving me the floor again. Paragraph 6. Regarding the chapeau, we welcome the articulation of the objective. at the beginning of the sentence, which clarifies that the objective is to uphold compliance with IHL. In the latter half, I will not repeat our thinking on the context-appropriate human judgment and control, but would like to register our reservation. On the other paragraphs, regarding paragraph 6, we support the first paragraph, but in the second sentence, I'm sorry, we support the first sentence, but on second sentence, like the distinguished delegation from France and Singapore have mentioned, we would like to delete ethical considerations. Assessment of legal obligations has clear basis, but ethical considerations does not have that clear basis. In addition, with regards to reference to effects of selection and engagement functions at the end of the sentence, we understood that this was intended to be in line with the characteristics of laws, which lists selection and engagement as essential functions, but we wonder whether it would be more reasonable to align with paragraph 4 of section 3 and replace it with the effects of attack. On paragraph C6D, Roman numeral 2, we agree with Spain that we should delete this, as it is unclear as to how we count the number of engagements. On Roman numeral 4, we concur with the view expressed by Spain and Singapore on the need to delete by nature as Article 52 of Additional Protocol I of Geneva Convention stipulates the possibility of other military objectives. Therefore, we propose to change the sentence to restrict the use of loss to military objectives. And as Singapore mentioned, we would be very curious to see, to know why by nature is retained. Lastly, on Roman numeral five, on the real-time machine learning, we think that the basis and rationale behind this is quite unclear. I would like to ask for its deletion. Thank you. Chair [2:36:13]: I thank the distinguished representative of Japan for sharing her views and making proposals. Next on my list is Canada, to be followed by Ireland and the Philippines. But first, the distinguished representative of Canada. You have the floor. Canada [2:36:28]: Thank you, Mr. Chair. My delegation wishes to weigh in on box number 3 for a second time with a focus on paragraph 6. As a general observation, Canada agrees with other delegations that the compliance measures listed in para. 6 represent specific measures to assist States in meeting their IHL obligations rather than applicable existing IHL prohibitions. The sub-paragraphs in para. 6 reflect standards and practices that apply to existing weapons with autonomous functions. it would be appropriate to include these recommendations in box 4, which prescribes the specific measures appropriate to laws. We believe the exception is para 6, which would be better suited in box 5. In paragraph 6, we find that the existing paragraph is somewhat ambiguous. We believe this paragraph is trying to reflect on measures that are deemed necessary to ensure that the effects and outcomes of laws and integrated technology components are consistent with the requirements of IHL. Specifically, the reference to predictability, reliability, traceability, and explainability are technical characteristics that are more usually associated with AI versus autonomy writ large. A better formulation might be as follows: ensure the effects of laws can be reliably anticipated and that the outcomes can be explained by those responsible for their development and use. We could otherwise support the alternative echoed by Australia to replace para 6A by para 4. In para 6B, We agree that this is an individual or personal related issue, issues that should be captured in box five. Moving to 6C, we agree with the principle that laws, mission perimeters must not be compatible, must not be capable of modification without, without human judgment and control, context appropriate human judgment and control, apologies. However, we feel this requirement would be better reflected as a sub paragraph to 6D, thus reducing repetition of controls, but still reflecting that both the setting of initial operating parameters and subsequent modification post activation both require human involvement. In summary, subparagraph 6C would be removed and in place a new 6D subparagraph should stipulate that ensuring that once activated, existing mission perimeters cannot be modified by the system without context appropriate human judgment and control. On 6D, I, we are okay. Likewise with 6D, Two, we are fine with the formulation as it is. On 6D3, we would propose to remove harm to civilians and replace with loss of civilian life, injury to civilians, or damage to civilian objects as to be more precise as to the nature of the harm to civilians. And finally, under 4, after "nature", we propose adding "location, purpose or use" to better reflect the wording of "military objective" in AP1. Thank you. Chair [2:40:26]: I thank the distinguished representative of Canada. Next on my list is Ireland, to be followed by the Philippines and Luxembourg. But let me first pass the floor to the distinguished representative of Ireland. You have the. Ireland [2:40:40]: I thank the chair for letting me take the floor. We welcome paragraph 6 and its subparagraphs. Some delegations have argued that context-appropriate human judgement and control lacks a shared understanding. The paragraphs clearly detail what restrictions are necessary to ensure context-appropriate human judgement and control and clarify any potential ambiguities. As noted in our comments on box 2, and in line with the interpretation of the ICRC and others, such control is not understood as direct or physical manipulation of the weapon system, but as control over its effects, brought about through a combination of prohibitions and restrictions that are further elaborated in box 3. We consider that that formulation helps to bridge divergent views and provides a basis for convergence by focusing discussions on practical measures rather than terminology alone. Additionally, many of the restrictions that are identified in the rolling text are reflective of existing military practice, which further demonstrates their feasibility and relevance. As such, it is essential that that section remains in box 3. as part of the two-tiered approach to regulating laws. Moving it to box 4 as a voluntary measure would defeat the purpose of that approach. In that vein, the word "should" in paragraph 6 should be changed to the word "must". Given that those are the steps that are necessary to ensure a system's compliance with IHL obligations, the word "should" is undoubtedly too weak. While generally supportive of the sub-items under paragraph 6, the concepts of predictability, reliability, traceability and explainability are essential for ensuring the appropriate control in order to comply with IHL as well as ensure accountability. In 6b, we welcome the reference to legal and ethical considerations performed by a human over the effects of the system. These reaffirm the central aspect of the human element, and for us it is important that it remains in that paragraph. However, the use of the term "chain of command and control" appears to be the combination of two terms, being "chain of command" and "command and control". A chain of command exercises command and control, and as a result we think that the phrase should be amended to read "operated under a responsible chain of command" or operated with an appropriate level of command and control. We also support 6 and its sub-items on clear limits. We would suggest in 6 to replace "reduce the likelihood of harm to civilians" with "avoid civilian harm". That would read: "limiting the operation of laws to a defined perimeter, in particular to avoid civilian harm or damage to civilian objects. Also, at para 6, the use of the term "limiting the operation of laws to a defined perimeter" does not seem to reflect the operational deployment of such weapons systems, in particular those that are airborne. We would suggest that the use of the phrase "operational control measures" would be more appropriate and would encompass a wider ambit of measures to control the operational deployment of laws. We welcome the updated rule 4 on the restriction of use of laws against military objectives, which we see as a key element. We thank the chair for. Allowing us to take the floor. for making comments and proposals. Chair [2:44:52]: Next on my list is the Philippines, and then I have Luxembourg, United States, Belgium, and the ICSC. Looking at the watch, I don't think we're going to make that, but we'll see where we cut off. But let me first pass the floor to the distinguished representative of the Philippines. You have the floor. Philippines [2:45:13]: I thank you, Chair, for this opportunity to share the views of our delegation on box 3. On paragraph one, we propose the following modification. It is prohibited in all circumstances to use laws which are of a nature to cause superfluous injury or unnecessary suffering, which are inherently indiscriminate, or which are not compliant with IHL. This proposal hopes to make the paragraph more concise, succinct, and direct to the point. Paragraphs one and two can also be combined into one paragraph. as they refer to the notion that the laws and its use should only be compliant with IHL. We also support the delegation of Singapore and their suggestion to replace "may only" with "must" in paragraph 2. On paragraph 4, our delegation views that the use of "anticipated and controlled" offers a clearer and more concrete requirement for the use of laws. We also appreciate the formulation of paragraph 5 and find it important to include design and development of laws in this paragraph. International law requires States to determine whether a weapon would be prohibited under international law under some or all circumstances when studying, developing, acquiring or adopting a new weapon. Our delegation appreciates the measures elaborated in paragraph 6 of box 3 as well as the attempt to operationalize context-appropriate human judgment and control. However, there is still a need to have a common understanding and greater clarity on this new terminology among delegations. Additionally, our delegation views that paragraph 6b's reference to ethical considerations is due to the mention of responsible chain of command and control in the first sentence. Hence, we interpret ethical considerations as referring to accountability in terms of the deployment and effects of laws. Nevertheless, our delegation welcomes further discussion on this matter if deemed needed. Thank you, Chair. Chair [2:47:15]: I thank the distinguished representative of the Philippines for sharing the views of her delegation. Next on my list is Luxembourg, to be followed by the US. But let me now pass the floor to the distinguished representative of Luxembourg, who is available. Luxembourg [2:47:36]: Thank you, Chair, for giving us the floor. Regarding the first five paragraphs of the third box, we have no further comments, and we support the way they are currently drafted. Thus, concerning paragraph 6, please allow us to make a few observations. First, regarding paragraph 6a, we propose replacing "I cite to those responsible for their use" with, I cite again, to those responsible within the chain of command and control, as stated under paragraph 6. We consider that this formulation more precisely identifies those who need to understand and predict the effects of laws. Second, regarding paragraph 5, we would prefer the term "requirements" instead of the term "prohibitions," as it refers not only to paragraphs 1, 2, and 4, but also to paragraph 3 that does not refer to a prohibition. Third, in regard to paragraph 6, we would also like to echo the comment made by France regarding the addition of the word "substantially" before the term "modified" in order to preserve an appropriate level of autonomy of the weapons systems. With regard to paragraph 6, we add our voice to the delegations that have underlined there is neither need nor reason to limit the number of engagements when they are carried out in accordance with IHL. Finally, regarding paragraph 60, roman 4, we would prefer, similar to Italy's proposal, to employ a phrasing closer to existing IHL as reflected in Article 52 of Additional Protocol I of the Geneva Conventions. Accordingly, we would suggest the following wording: "Restrict the use of laws against military objectives as defined in Article 52 of Additional Protocol I of the Geneva Conventions." Thank you for your patience, Chair. Chair [2:49:51]: Let me thank the delegation of Luxembourg for stating their position. Next on my list is the US, and I still have Belgium and the ICSC. But let me first pass the floor to the distinguished representative of the US. United States of America [2:50:08]: Mr. Chair, thank you for giving my delegation the opportunity to provide specific comments on paragraph 6. First, regarding subparagraph A, Predictable and reliable need to be tied more to specific IHL rules. IHL does not establish general standards for reliability or predictability of weapons. Predictability and reliability in relation to effects on civilians are aspects of the prohibition on inherently indiscriminate weapons and the prohibition on attacks expected to cause excessive civilian harm. So although we cannot support These as freestanding requirements, we are open to working on these concepts to give them more granularity and a connection to existing IHL rules and requirements. Regarding traceable and explainable, these are relatively new terms used in the context of discussions on AI developments without well-developed common understandings. These are not requirements of IHL, and laws do not necessarily involve the use of AI capabilities. Indeed, our characterization of laws does not indicate that AI capabilities are an essential element. Regarding paragraph B, we support the first sentence, but see as a measure appropriate in box 5. Like Israel and Singapore, we have concerns with the second sentence of subparagraph B. We do not see the second sentence of subparagraph B as emanating from the first sentence, nor as specific enough or tied to a rule of IHL to be effectively implemented in practice. Regarding subparagraph C, we do not think this statement reflects an accurate technical characterization of these systems. Autonomous systems are only capable of performing functions they are designed, engineered, and enabled to perform. If a system can move to a geographic area outside of the immediate environment where it is deployed, all that means is that the developer engineered the system and the operators enabled it to operate in that wider environment. An operator might fail to properly determine or specify operational parameters, for example, if they lack training or a proper understanding of the system, but that does not mean the system modified its own parameters on its own accord. If this paragraph is meant to describe a purely hypothetical capacity that is yet to be demonstrated, we do not think it is appropriate to include in an instrument. Regarding subparagraph D, first, some of these specific measures are not done in the use of the system, which is the chapeau of this paragraph. The factors mentioned in this paragraph can be set or implemented during development, and it may not be possible to implement them during use. Regarding subparagraph D, Roman IV, existing IHL does not have this restriction, and we are concerned it could encourage bad actors to use more ostensibly civilian objects for military purposes. regarding subparagraph D, Roman 5, we support the comments of the Republic of Korea and the request of Japan to delete this subparagraph. I thank you, Mr. Chair. Chair [2:53:45]: I thank the distinguished representative of the United States. Next on my list is Belgium to be followed by the ICSC, but let me first pass the floor to the distinguished representative of Belgium. You have the floor. Belgium [2:54:01]: Mr. Chairperson, in Belgium's views, overall, the text in this box seems to be in good shape. In general terms, this box contains the elements we are looking for. We view this box as a key section for the incorporation of the two-tier approach into the text. At this stage, our comments will be limited to paragraphs 1 to 5. Listening to the interventions made in the room, We can echo the comments made by some previous delegations regarding paragraph 3. We would like to underline that we understand that the words "civilian objects" refers to civilian objects which may not be qualified as military targets, and that "individual civilians" refers to those individuals, unless, as for such time, as they do not take a direct part in hostilities. We support the previous comments asking to add the prohibition of attack against other protected persons and objects under IHL under this paragraph. Regarding paragraph 5, we welcome that the prohibitions included under paragraph 1 to 4 are to be taken into account during the design and development phases of the life cycle. However, we think that we could go even further by including the development phase in the explicit prohibition under paragraph 1. This also reflects the idea included in the current paragraph 9 of box 4, stating that states undertake not to manufacture laws that cannot be used in compliance with IHL. We, therefore, suggest to change paragraph 1 to read as follows: It is prohibited in all circumstances to develop and use laws which are of a nature to cause superfluous injury or unnecessary suffering. In such a case, paragraph 5 could be deleted as it becomes obsolete. I thank you. Chair [2:56:12]: I thank the distinguished representative of Belgium for sharing the views of his delegation. And we have less than two minutes left. And I think a number of delegations have still reserved a position with regard to box 6. So I do not think we can close the discussion on box number 3. So I would like to start tomorrow then with the ICSC, if that's OK with you. I will also thank the interpreters again for an afternoon of hard work, a day of hard work. I'm not trying, I'm not going to try and summarize the discussion from today. Let me reflect a bit on it. Sometimes I would almost be tempted to say that a bit of machine learning might help me then to do it overnight. But I think we had a very good discussion and I'm also grateful that you were really very interactive also reacting to the proposal of other delegations. And also explaining your own positions, not just stating, uh, why or why not you like a certain paragraph, and that's been helpful. Um, we will continue our discussion tomorrow, uh, on box three, and I intend also to go through box four and five, as I stated, uh, on Monday, that we have done a full reading by the end of Wednesday, so you know what will be our target for tomorrow, in the course of tomorrow, I will then also share with you. what are my intentions with regards to Thursday and Friday. But first, we have to go through the motions of the full text. Thank you very much for that. I wish you a pleasant evening and see you tomorrow morning at 10:00 again in this room. Thank you very much. The meeting stands adjourned.