8th Meeting - 1st Session Group of Governmental Experts on Lethal Autonomous Weapons Systems 2025

Okay, thank you very much. Excellencies, distinguished delegates, this morning we went some, we discussed some text suggestions with regard to box one. I think we had a very, very thorough discussion, but I already announced that there will be an opportunity in the afternoon to share any brilliant after-lunch thoughts with the other delegations. But if there is no further appetite for continuing the discussion, because we had indeed a very thorough discussion in the morning on box 1, I think we have a lot of takeaways from that for further improving the text and for coming up with suggestions that contribute to narrowing down still existing divergences. I stop here because I recognize Spain, so I first give the floor to the distinguished delegate from Spain.

Thank you, Chair, and I apologize for taking the floor this afternoon. I couldn't participate this morning. We thank you once again for your work, and we thank your team as well for presenting draft With a view of achieving consensus, I'd like to make some comments on behalf of my delegation on paragraph 1, your proposal to include.

In the scope of the application of the CCW.

This is very appropriate and adequate. As stated by other delegations, for example, Italy and Germany, we would thank you for including such concepts as once activated or without human intervention. On paragraph 2, we support the proposals made by France and Sweden to merge 1 and 2, because we believe that this should be part of the characterization of laws. Thank you.

I would like to thank the distinguished representative from Spain for sharing her comments and making proposals. Is there any other delegation who would like to engage on box one? If that's not the case, then I would really like to thank you for a very thorough discussion this morning. As I said before, I have lots of things to, uh, to, to think about, uh, with regard to proposals made, but I think we have a, a number of constructive proposals that could facilitate, um, um, the way forward with the rolling text, um, the, we will, as a team, we will look into that, um, um. when developing a new version of the text. So thank you all for that. I think it has been most useful to have this feedback on the possible ways forward with certain elements. As I indicated earlier this morning, there are, of course, still some other issues that we have to solve. There will be time, of course, for that also in the intercessional period, and I will make some announcement with regard to that tomorrow at the end of the afternoon. Then now let's move into box number two. Perhaps could you put up the text of box number two? Yes, thank you for that. Box two as a section has appeared to garner a relatively wide support from the delegations of the group. And I can say that most delegations, I think, welcome the fact that the text reflects language from previous reports adopted by the group. and of the preamble part of the CCW. There's overall broad, but not yet unanimous support with regard to referring to development and use of laws throughout the text, so we will have to do some fine tuning on that. But we have, of course, also received some proposals for mending some of its paragraphs, both from the floor and in written form. My proposal is to retain the present formulation of para. 1, since it has received wide support from delegations. With regard to para. 2, I would like to suggest the replacement of "independent" with "regardless" in order to capture more clearly the intent of this para. And I would also like to propose the omission of the term "military" in para. 2. in order not to narrow down the scope of the application of IHL unnecessarily and to avoid any loophole in this provision. Then with regard to para 3, that also remains the same for now, as its formulation makes clear that it is a logical sequence from the para that precedes it. Para 4, On para 4, we received a number of comments, and I have sought to accommodate the varying views of delegations. For streamlining purposes, I propose that we remove the first part of the para, but insert "with respect to laws" in order to be clear that para 4 relates to laws as the subject matter that we are working on. Para 5 remains as it is for now, and I have not yet suggested a modification on this para, as I would like further discussions on this matter, especially on how delegations understand context-appropriate human control and judgment. We had already quite some discussion on that, but I do not feel comfortable yet to make any proposals with regard to changing the text, and this is, of course, a topic that we cannot avoid, we have to address it. And we'll come back on other places also in the text. So it's still there, and we will have to work on that later. And I think later, I mean today, because I would also like to ask delegations to specify how exactly you understand context-appropriate human control and judgment. how we can find a solution, a formulation that is in the end palatable to all of you. And let me put some questions on the table, um, to, to steer in a way the discussion on, on, on human control or control or control by an operator. We have seen many, many versions. Question number one, is direct control or oversight during identification, selection, and engagement stages always needed to comply with IHL, or can we agree that by using the term context appropriate human control and judgment, we acknowledge that in certain systems, human control and judgment can be exercised before identifying, selecting, and engaging a target? Two, Does human control entail judgment or vice versa? Or do we agree both cover essential concepts or precautions to ensure compliance with IHL? And that brings me then to the question, at what stage is human control crucial? Third question, to what extent does human control take away the autonomy of a weapon system or disqualify a system from being labeled a loss, and in what stages after its activation? And the last, what specific contextual factors should be taken into account when deciding on the necessary amount of human control and judgment in a specific case or with regard to a specific system? Is this linked to external factors, like areas with a dense population? or civilian objects. The reason for the questions is to allow for a deeper understanding why we have the divergences in the room that we have at this moment, and to give delegations a better understanding of the different positions, and hopefully give me some guidance with regards to a possi- to finding a possible landing zone for consensus. Having said that, um, if you would like to discuss also other parts of the text, you are of course most welcome. We have the text as such on the screen. Let me now open the floor and… We are all hoping for somebody to be as brave to, to have some opening remarks with regard to these questions. Thank you. And of course, you could also react to the changes proposed in the other paragraphs because this is in fact with regard to paragraph 5, so let's be honest. I recognize the Distinguished Representative of the USAs. Sir, you have the floor, and thank you for breaking the ice.

Mr. Chair, thank you for the opportunity to talk and to you and your team for all your intensive efforts on this section to improve it. We greatly appreciate that. Also really appreciate the very rich discussion and insightful contributions from colleagues this morning. We're listening to those and really trying to understand delegations' positions and the issues. I'll just try and briefly address, you know, the first several paragraphs, putting aside your questions, which we're still trying to digest and prepare some thoughts on. But just with regard to the first number of paragraphs that you've already circulated, just wanted to give a couple comments. With regard to the second paragraph, IHL applies in armed conflict, whether international or non-international and governs the use of all weapons, means, and methods of warfare, those of the past, those of the present, and those of the future. I think we fully support that part of the sentence. I think we appreciate how you've refined this last clause, and is consequently applicable independent of the technology used. Or pardon me, regardless of the technology used, I think we can support this change from regard from independent to regardless, we think that's a very positive change, we also support this deletion of military, we think that makes a lot of sense to us. That said, for us, the application of IHL rules can depend on the specific technology used. And here I'm thinking of things like the CCW protocols themselves, that if an incendiary technology is used in the weapon, then it can then fall under the CCW Protocol 3. Or if a blinding laser technology is used in the weapon, then particular IHL rules do come into play. And so we do see some, especially in the area weapons laws, some ways that IHL's application does in fact depend on the specific technology that is used. That said, we agree with the formulation used in the International Court of Justice's advisory opinion that the cardinal principles of IHL would apply regardless of the military technology used. We also think that the rules on attacks, the rules governing attacks, that those rules, important principles and requirements of distinction, proportionality, precautions in attack, A number of other rules, warnings, et cetera, the prohibition on attacking undefended cities, right? These rules apply, just to borrow the formulation in the Hague regulations and in the First Additional Protocol, those rules apply by whatever means is used to conduct the attack. And for us, that seems to be the most pertinent proposition in our discussion on laws, that we believe laws are weapons. They're lethal weapons used to conduct attacks. And regardless of whatever means is used to conduct the attack, whether it's attacking from a balloon, attacking by a bombardment, attacking through an assault, The core principles and requirements of distinction, proportionality, and precautions of attack continue to apply. Therefore, our proposal would be to omit the clause and is consequently applicable independent of the technology used and replace it with the following, including rules governing attacks that apply regardless of the technology used to conduct the attack. So I'll just read it out again for colleagues. IHL, it applies to armed conflict, whether international or non-international, governs the use of all weapons, means, and methods of warfare, those of the past, those of the present, and those of the future, including rules governing attacks that apply regardless of the technology used to conduct the attack. And we think this would also really be pertinent in the context of laws because we believe these are weapons that are used to conduct attacks. So that's our comment, our analysis with regard to that paragraph and our specific suggestion to try and move us towards consensus We did also want to make a comment with regard to paragraph 4. I think we appreciate this notion of directing our attention to laws. And I think we've made a number of proposals in the document to focus on how can we link our discussion of IHL principles and requirements, how can we link these measures to be very specific to laws. So we're very aligned with that objective. That said, I think our suggestion as far as an approach that would, we believe, attract the greatest consensus would be to use the formulation of the Martens Clause that's used in the CCW preamble. And I think one point of rationale here is, I think there could be other ways of framing it, I think the mere fact that it's placed in this instrument could give an indication of our connection and focus on laws. I also think that, for example, the first paragraph that we have also comes from the CCW preamble, and we think it's relevant and pertinent, but there isn't a specific reference to laws. So I think that would be our specific proposal, just to go back to the language, "in cases not covered by the CCW and its annex protocols or by other international agreements," and then pick up the language, "the civilian population and combatants." So, so that's, those are our suggestions with regard to the, the first four paragraphs. You know, I think one, two other ideas that we would offer for your consideration, Chair, as well as the consideration of other delegations. One is, is thinking about this section as kind of a preambular section for our elements. That's how we've been thinking of it. And, and Of course, as a sequencing matter, the preamble usually goes at the beginning of the document. So that's one question to think about that we'll be continuing to think about. And then another question for us was whether the Martin's clause paragraph should go in a different place in the section. Again, we don't have a strong view or position, but really did want to note that as an area we're continuing to think about as we're trying to be as constructive as possible and to try to make this the best possible set of elements of instrument. I thank you, Mr. Chair.

I would like to thank the distinguished representative of the US for sharing the views of his delegation and for providing us with suggestions. With regard to the Martus Kosia, I would say it's either at the very beginning or at the very end. It's at the very beginning because this is what there is before there's something new, or at the end, whatever is above, this is always valid. But I'm open to other suggestions, of course. and perhaps this might trigger them. I have on my list Belgium, Australia, and Switzerland. But let me first pass the floor to the distinguished representative of Belgium. You have the floor, sir.

Thank you, Chair. My delegation will need a bit more time to reflect on the questions you posed earlier to be able to give you a specific answer to them. However, we would like to take this opportunity to express our gratitude to you and your team for the efforts on box 2. We were already quite pleased with your original proposal for box 2, and we can also support the revisions you proposed yesterday. We have one suggestion regarding paragraph 3. Our proposal would be to rephrase to international law, in particular IHL.

So it would read.

International law, in particular IHL, therefore continues to apply fully to the development and use of laws. Thank you, Chair.

Let me thank the distinguished representative of Belgium for sharing the views of his delegation and for providing us with some proposals. We will look into whether we later on could get, might get the questions on the screen, but that's for later. Let's first go through the paras and then see whether that's possible or not. Next on my list is Australia, to be followed by Switzerland. But let me first pass the floor to the distinguished representative of Australia. You have the floor.

Chair, we thank you and your team for all of your efforts in preparing this revised text, and wish to make a few observations in the revised drafting of paragraph 4. Chair, if the Martens Clause is to be restated here, we would prefer to align it more closely with the wording included in the preamble of the CCW. Using the current formulation on the screen, there is a risk that it could be interpreted to mean that international agreements, such as the Geneva Conventions, do not apply to laws, which we don't think is the intended effect. I'll explain this further. So the Martens Clause, as expressed in the preamble to the CCW, applies in cases not covered by the CCW and its annex protocols or by other international agreements. In such cases, the civilian population and combatants shall at all times remain under the protection and authority of the principles of international law derived from established custom, from the principles of humanity, and from the dictates of public conscience. But by saying that such principles apply with respect to laws without the qualifiers included in the original text, there is a risk that this could be interpreted to mean that other international agreements, such as the Geneva Conventions, do not apply to laws, or in other words, that laws are cases not covered by the CCW and its annexed protocols or by other international agreements. We would therefore suggest reverting to the previous language that you had included in the rolling text, which closely replicates the wording included in the preamble to the CCW and avoids unnecessary ambiguity. Thank you, Chair.

Let me thank the distinguished representative of Australia for sharing the views of her delegation and for proposing changes to the text. Next on my list is Switzerland, to be followed by the Russian Federation, Cuba, and the Asia-Pacific Institute. But first, I would like to give the floor to the distinguished representative of Switzerland. You have the floor.

Thank you, Mr. Chair. simply repeat what we said in the previous statements on this box. We are generally very happy with it. Paragraphs 1 to 4 are basic and fundamental restatements of IHL or aspects of the general principles. That said, we have A similar concern, as was very eloquently expressed by our colleagues from the USA and Australia just right now, with regard to paragraph 4. We understand the purpose of the Martens Clause to specifically remind that in cases where an instrument does not cover certain aspects, for instance, certain means or methods of warfare, civilians and competence continue to be protected and that the conduct of belligerents remains regulated at the minimum by the principles of international law, principles of humanity and from the dictates of public conscience. So we share the view that it would be better to stick to the formulations usually used in other instruments, including in the preamble of the CCW, that is to quote In cases not covered by, and here we could use instrument for now, or by other international agreements, the civilian population and combatants remain, and so on. Of course, instrument would then need to be clarified once we know the form of this instrument we are discussing here. Thank you so much.

Let me thank the distinguished representative from Switzerland for sharing the views of his delegation. Next on my list is the Russian Federation, followed by Cuba. The distinguished representative of the Russian Federation, you have the floor.

Thank you very much, distinguished Chair. Overall, my delegation does not have any significant difficulties with the language that is being proposed by you in box 2. We find that the language and the amendments that you introduced do indeed reflect our own position and help us achieve results based on compromise. On the comment that was made by our colleagues from the United States of America about paragraph 2 and the ending of the paragraph, and we have to say that paragraphs 2 and 3 in their current version is something that we are fully happy with. We could in fact think about merging paragraphs 2 and 3. in order to alleviate the concerns and the difficulties that the delegation of the United States has with the last part of paragraph 2. Instead of the language that we have here, we can simply use the language we have in paragraph 3 that is very comprehensive and it covers the issue of the applicability of international humanitarian law in full. related to the development and use of laws. Perhaps that could be the solution. On paragraph four, the language that you are proposing, Chair, is something that we're satisfied with. On one hand, And this language creates a link to the subject that we're discussing here. And in other words, weapon systems based on new technology in the area of laws. And on the other hand, it is of comprehensive nature. And we believe in this way, it does indeed cover cases that fall under the CCW and also it and other international agreements, and also cases that are today not covered by the Convention due to the language that is of general nature. On paragraph 5 and the notion of human control and judgment, as we understand, although this will require some additional and very serious consideration, we would like to stress and once again reassure you that we're ready to be flexible as far as this notion is concerned in order to find compromise-based solutions. However, for us it is clear that this term should be a simple one. a term that is something that we understand in order for all of us to be clear what it means. The use of such a long term that is pretty vague, something that we have right now, is something that we don't find to be advisable and we prefer to have it shortened and limited only to human control, something that we would prefer, because it is indeed a term that is easily understood. Or we would be ready to look at any alternative language that other delegations are proposing as they play around with the words control and judgment. Thank you.

Let me thank the distinguished representative of the Russian Federation for giving feedback on the proposals made. Next on my list is the distinguished representative of Cuba. Madam, you have the floor.

Thank you, Chairman. Just two comments, just on word drafting generally. editorial comments on paragraph 3. We could think about the possibility of saying that international law fully applies to laws, including in their development and use. If we mean that development and use are stages via which laws pass through otherwise we're leaving it outside the applicability of IHL as a as stages such as destruction and design because uh and development so Therefore, we should maybe reformulate the structure of the wording just so that we understand fully that these are part of the systems or the life of the system. And then power for perhaps. Seek, we need to seek some way with respect to laws as you propose. Lee. linking it to protocols from the relevant international relevant agreements that would also give us margin to have the paragraphs for that we assign for applicability to IHL also I agree with the other And other than that, I agree with the proposals made so far and the way that we're working on these paragraphs. Thank you.

Let me thank the distinguished representative of Cuba for sharing the views of her delegation. Next on my list is the Asia Pacific Institute to be followed by Canada. But first now, the Asia Pacific Institute. You have the floor.

Thank you, Chair. We would also appreciate some more time to digest your questions and seeing them on the screen would be helpful. But meanwhile, we can offer a few comments on other aspects of the text. In paragraph two, the replacement of the word independent with regardless and the deletion of the word military are welcome improvements. This paragraph now better captures the idea that the body of IHL applies irrespective or without prejudice to the technology used. And this proposition seems to be consistent with the US observation that some rules of IHL may relate to specific types of weapons or specific uses of weapons. Now, paragraph 3 seems to follow directly from paragraph 2, which is even emphasized by the use of the word "therefore." And to further emphasize this point, it might be textually appropriate to merge paragraphs 2 and 3. Chair, the changes to paragraph 4 reflect a commendable attempt to simplify the text. However, the new text may have inadvertently created additional complications, which Australia and Switzerland have already alluded to. As many delegations have previously pointed out, this paragraph reflects the Martens clause, which can be found in many IHL instruments since 1899, including the CCW. There are different views as to the contemporary significance of the clause. However, the fundamental purpose of the Martens clause is to create a legal safety net for situations where states have been unable to agree upon specific rules. It may also be relevant in situations where the application of existing law is uncertain. In the version of the Martens Clause found in the CCW preamble, this gap-filling function is made clear by the use of the words "in cases not covered by this Convention and its annexed protocols or by other international agreements." Removing these words would alter the meaning of the CCW preamble. Furthermore, the proposed new clause with respect to laws is ambiguous in terms of its scope and creates novel issues of interpretation. Like Australia, we therefore suggest returning to the Chair's original language. For clarity, we also suggest making it clear that the cases involving laws refer to cases involving the use of laws. The paragraph would then read, "In cases involving the use of laws not covered by the CCW," and so on. I thank you, Chair.

Let me thank the distinguished representative of the Asia Pacific Institute for his comments. Next on my list is Canada. You have the floor.

Thank you, Mr. Chair. I can be very brief. Firstly, my delegation wishes to thank you and your team for the work that you've undertaken in trying to seek a pathway forward. On section 2, paragraph 4, just to note that my delegation agrees with the suggestions that have been made by Australia. We see value in retaining the CCW language in cases not covered by CCW and its annex protocols or by other international agreements, the civilian population, and so on, as we have written there. I thank you.

Thank you for that. And next on my list is France. Let me pass the floor to the distinguished representative of France. You have the floor.

Thank you, Chairman. Very quickly, like others who've spoken before with much greater eloquence than me, we prefer to use the language of the CCW preamble in paragraph four. The rest of that box two, we're fine with it, and thank you again for all your efforts for, in proposing the new text. Thank you.

Um, next, next on my list is the distinguished representative of Belarus. You have the floor, sir.

Thank you, Chairman. Very quickly. In para 4, the words "civilian population" and "competence" could maybe replace it with "all people", "all persons". Again, it repeats the CCW preamble, but strictly speaking, it's not entirely consistent with the system of the four Geneva Conventions, because wounded servicemen are not formally civilians, and at the same time, they do fall under the protection of IHL. And effectively here, we're talking about everyone, essentially talking about all persons. So it's just a small technical suggestion. Thank you.

Thank you for that technical suggestion. Yes, I recognize the distinguished representative of Pakistan. You have the floor, sir.

Thank you, Chair. Great appreciation to you and your team on producing this revised text in box 2. At this moment, we would just like to briefly comment on paragraph 4 and the discussions that have followed here on this. We do see value on retaining the language that is contained in the preamble of CCW, but as pointed out by the delegate from Switzerland as well, we have not yet made determination whether this is going to be a protocol, whether this is going to be any instrument. Of course, if there is convergence and everybody in this room agrees, which we would prefer as well, that this is going to be a protocol under the CCW, we can use the same language that in cases not covered by the CCW and its annex protocols. But for now, I think what makes more sense is in cases not covered in this instrument, and then we go on with the civilian population and the combatants. We are certainly not in favor of inclusion of word with respect to laws, simply because we have not made that determination yet, whether there will be certain laws which will not be covered in this instrument. And I think that's a discussion still taking place in the context of box one. So for now, we can just say, as proposed by the delegate of Sudan, that in cases not covered in this instrument. Thank you.

I understand that we are mastering the technology and we are going to put up the questions on the screen. Just to be sure, they are just to assist the discussion. And as I said before, we would also appreciate from your side if you could illustrate certain proposals also with examples. I have now two requests for the floor, Israel and Austria. Let me first pass the floor to the distinguished representative of Israel. You have the floor.

Thank you so much, Chair. So we will proceed to give our comments on section 2. Chair, at the outset, and following some remarks heard this morning in the room, we stress that we maintain a professional approach to the discussions, respect all viewpoints expressed, and believe that this is the way these discussions shall be maintained with regard to the perspectives of all delegations. Now, with regards to section 2. Regarding paragraph 2, we support all the suggestions made by the American delegation. Regarding paragraph 4, we are restating the position we presented a few days ago. We, like other delegations, think that the original text of this paragraph should be maintained, given its status as an historical agreed language. and the need to avoid implying that the Martens Clause means something different than it has been meant so far. We are struggling to see the value that the removal of the first sentence provides that is worth the risk of creating a new and unclear meaning for this historic text. Finally, and importantly, we want to express again our objection to the language "context-appropriate human control and judgment" The term human control is highly controversial and was never adopted in any consensual outcome as the appropriate description of human machine interaction for a good reason. Our opinion is that the term is its substantive problems has remained the same. We see a serious problem also with the phrase context appropriate. that can easily be understood as referring to an obligation of constant real-time control, which is in the heart of these debates. For these reasons, we request using a language that has already achieved consensus, such as appropriate human involvement. We have heard this week other alternative phrases, that we are open to them as well. We also recommend placing the paragraph in later sections on this paper, since The current section is focused on fundamental issues concerning the applicability of IHL, rather than IHL obligations themselves. There are lone elements which are not even existing obligations. Thank you, Chair.

Microphone, please.

Let me thank the distinguished representative of Israel for providing us with his views and comments. Next on my list is Austria. You have the floor.

Madam Chair, and also from our side, thank you for producing this text. I think we haven't said it in box 1, but in box 2, thank you to your team and all the hard work that you put in last night and today. We will also focus on box 2. We are still thinking very hard on the questions, and it will take us some time. But just on the discussions that we are currently having, I think we share the understanding that we would kind of change the character and the idea behind the Martins clause when we put it onto the laws debate. And it would kind of be an expansion of the Martins clause into a lower level by explaining that it also is applicable to laws. But I think that was not the, the original idea here. It's a good point to reaffirm the Martens clause here because it relates to, to new technology and it also has, I mean, next to the catch all function that it, that it has, it also has another idea built in, which is that there is a normative normative idea behind IHL that it should be complete and that there is nothing in the armed conflict that is not governed by something. And maybe this is the idea that you wanted to capture. And here, just when I looked up the Martens Clause as it's formulated in the Preambula, I just saw that there is also another part in the Preambula which also captures this idea, this normative idea, to have completeness and development. of IHL or law applicable to armed conflict, which is the one reaffirming the need to continue the codification and progressive development of the rules of international law applicable in armed conflict. And maybe this would also be a part of the CCW preamble that it would be useful here. I mean, it would be more for decoration and actually, um, for the, for the, for the, for the, for the, for doing something, but it would give it another nice touch and maybe capture some of the original idea that was here when the Martens Clause was linked to laws. We would also prefer the most broad formulation of the Martens Clause, which would be in cases not covered by this convention or the CCW and its annex protocols, that would be our preferences. Uh, coming shortly to five, I mean, um, our position is that meaningful would be a, a very good addition here, um, but we know that this is, uh, uh, the current text is the one that has a strong preference, uh, but still to, to keep this in mind that there is also another possibility to enhance or even improve the text. When it comes to why this part is here, we see it as the other parts in this section as something that reaffirms IHL. And it's, as explained earlier this week, it is an idea that has been formulated in previous texts, in reports, and papers produced by this body. And it's more or less an expansion or development of this idea that we have already formulated, how our work or how human control relates to IHL, and that's why we still support the inclusion here and not in box three. Uh, that was just another point that we wanted to make. We are still looking into the suggestion on para two. I think the current formulation works, but we are looking into the, the suggestions that we have received from the United States, too. Thank you.

Let me thank the distinguished representative from Austria for its comments, suggestions. Next on my list is Germany. You have the floor, madam.

Mr. Chair, Germany welcomes the new version of part two of the rolling text. Given the inherent dual use character of the technologies under discussion, we particularly appreciate.

The deletion of the qualifier military before technologies.

Germany can go along with a new version of para four. At the same time, we agree with Australia, Switzerland, the US, Canada, and France that there may be merit in reflecting the Martin's clause more precisely as it was the case in the original version.

The suggestion made by Switzerland and Pakistan in order to reflect the wording of the CCW Preambula more precisely may.

Also be a valuable option that we can look into. We would be flexible with regard to moving parts of the para to another part of the document. Thank you.

Let me thank the distinguished representative of Germany for her statement and comments. brings me at the end of the list at this moment. Assuming Austria is still an old plaque. We have put up the questions on the screen. They are not a purpose in itself, of course. They are just to stimulate the discussion and to promote a better understanding Some delegations have already shared their views on it, but also for other delegations. For me, it's just to better understand the different positions with regard to what we have, with regard to the comments so far received. We will again look into the beginning of para 4. There were some suggestions made. We are going a bit up and down, but I think that We can find a fix for that. I see one delegation already volunteering to kick off. That's great, but let me give you a few more minutes to have some internal discussions so we can have a proper follow-up and you don't have to still work on your speaking notes and listen to colleagues. 35 and then we kick off. Excellencies, distinguished delegates, shall we resume? Earlier, I saw the plaque of Pakistan. Is that still valid or was that...? Good, great. You have the floor.

Thank you, Chair.

Please, please, delegates. We have a volunteer to kick off the discussion, so I think we owe him some respect and attention. The floor is for the distinguished delegate from Pakistan.

Thank you. I believe now you are inviting some preliminary views on some of the questions that are on the screen, and I would like to take this opportunity to also share our general comments on the paragraph five and specifically the terms of context appropriate human control and judgment and some of the views that we have heard in the past few days and today as well. First, we would like to recall very clearly what we have said previously, that the core reason for introducing the concept of human control in the first place is precisely to address the unique challenges posed by autonomy. As we noted before, some had earlier pointed out that the phrase is not present in the existing IHL text. Yet, as we believe, the absence of explicit language in historical treaties, conventions is not an evidence of its irrelevance, but it highlights the unprecedented nature of the questions that we are asked to address in the course of our work. Until now, we believe that human control has always been assumed intrinsic to all past weapon systems and technologies. Regarding your question, how human control is linked with human judgment, we believe that they are interlinked. They're linked to one another. We consider them very close to each other. But human judgment provides the critical reasoning and ethical considerations required by IHL, while human control ensures that this judgment translates effectively into lawful action. So judgment, in our view, without the possibility of exercising control, is meaningless, while control without informed human judgment is also inadequate. So we see value in retaining both of the terms as we have tried doing so in the past. For instance, in the work of 2019, we agreed on how judgment is essential for ensuring compliance with IHL. And in 2023 GG's report, we made a similar conclusion in the context of control. Now, we have raised, we have listened to some concerns in the past regarding the term meaningful human control. What we have heard is that it's very vague or potentially very narrow. And it is precisely to address these concerns that you introduced this phrase of context appropriate human control and judgment. And we believe the term offers that flexibility, acknowledging that the extent and form of human involvement necessary for compliance with IHL will vary naturally based on the operational context technology and scenario specific demands. And here lies the answer to your questions where you invite that what are those context specific factors. And in our view, it's not going to be one exhaustive list. Of course, there is operational environment, there is complexity, clarity of the targeting decisions, there are capabilities and limitations of certain systems, the nature and duration of mission as well. So in our view, this phrase offers-- it's not an attempt to constrain states or impose rigid requirements. Rather, it serves to clearly link the necessity of human oversight directly to compliance of IHL principles. And it represents-- it continues to represent the best available compromise on these issues, one that is flexible and addresses practical concerns. in the case of laws. We are also open to the idea of reflecting these two terms separately as well in this context as we have done in the past to talk about judgment in a separate sentence and context appropriate human control in separate. We remain flexible on that as well. But we continue to believe that box 2 is the place to talk about these concepts. Why? Because we are not just restating the principles of IHL here, but we are listing all those important guiding preambular principles in the case of this instrument. And the context-appropriate human control and judgment, in our view, remains one of those guiding principles which should help us define and lay down those parameters in the subsequent boxes. Thank you.

Let me thank the distinguished representative of Pakistan for sharing his views with regard to the guiding questions or whatever you want to call it, at least the questions to help us forward with regard to the text we have here. Next on my list is the US. Let me pass the floor to the distinguished representative of the US. Sir, you have the floor.

Thank you, Chair. So not that you need to know this much information about our delegation, but I think there are some on our delegation that prefer classical music, and there are some on our delegation that like jazz, which is a little bit more improvisational. And so we've been having a debate about whether we're approaching this from one perspective or the other perspective. Just given the questions and the amount of time that we've had, I think this is going to be more jazz than classical, meaning there's going to be some improvisation and we may miss some notes. So just to try to react to these questions, on the question of to what extent does human control take away the autonomy of a weapon system or disqualify a system from being labeled a laws and in what stages after its activation, So I think from our perspective, this is a little bit hard to categorize in terms of what question are we trying to answer and how is it relevant here versus relevant to say other parts of the document. I think if the operator is controlling the essential functions of a laws like the selection and engagement of the target when operating the system. then in that context, the system is not being operated as a LOS. But that isn't necessarily the important question because the fact that a LOS may be capable of being operated in a manner that would not use the autonomous features to select and engage doesn't change it from being a laws if the system has the ability to carry out those functions after activation without further intervention by an operator. So we're trying to answer the question And I think it helps us get to a deeper level of understanding. But we're not sure that it illustrates the need for human control or some degree of human control. Similarly, the question about the specific contextual factors that should be taken into account when deciding on the necessary amount of human control and judgment in a specific case or with regard to a specific system, is this linked to external factors? So, as we just noted, we don't believe it's necessary for a human operator to control the weapon when it is autonomously performing these critical functions. So, for our delegation, I'm not sure there are contextual factors that would need to be taken into account in determining the amount of human control that is necessary. But the question does illustrate the importance of the commander and operator exercising judgment. And so maybe this is one way to understand the difference between control and judgment, that it is important for a commander and an operator to exercise judgment in considering whether and how to use a laws depending on a variety of contextual factors. including whether to deploy the laws in the first place if it's going to be used against a military objective. I think some of these factors are illustrated in boxes three and four when we're trying to sort of get at the kinds of precautions, for example, that might be feasible in considering when and how to use the laws. Those are things that would require the judgment of the commander or the judgment of the operator. As our draft articles proposal indicates, To ensure effective implementation of the principle of distinction in attacks when using the laws, commanders and operators must have the intention of striking specific or potential targets that constitute military objectives, or of operating the system within specific locations constituting military objectives to deny enemy forces access to such locations. So those are examples of judgment that is needed by an operator or by a commander when thinking about whether and how to use a laws. So we're doing our best to answer the questions, but I do think that there is a degree to which we're not sure that we agree with the premise of some of these questions, which is that there is some amount or degree of human control that is necessary to ensure compliance with IHL when using a laws. As noted by the delegation of Israel, This is a term that has never achieved consensus in this GGE, and from our perspective, that's for good reason. It's a vague formulation. It doesn't have an anchor in international law, and it is clearly susceptible to multiple and different interpretations. But more fundamentally, and as some of what I've already said indicates, there are many examples of existing weapon systems that perform the essential functions of selection and engagements with no human control. And there is no question that these systems can be operated in compliance with IHL. So for our delegation, this is a direct recognition that autonomous control of these functions can be sufficient to ensure compliance with IHL. The distinguished delegate of Pakistan noted that we grappled with this in the context of the 2023 GGE report, and we had very, very good discussions at that time about these very questions. We have been discussing, debating these issues for many years, and the quality of discussion on these questions is very, very high. But I do think it's important to recognize that in paragraph 21 in the 2023 report, we don't refer to human control. We refer to control. Control with regard to weapons systems based on emerging technologies in the areas of laws is needed to uphold compliance with international law, in particular IHL. And that, in our delegation's view, was a recognition that for many delegations, It was important to leave open what NAR delegation's view is the correct view that control with regard to weapons systems can be human control. I think we've always said that that is an example of the type of control that can be exercised to ensure compliance, but it's not the exclusive or the only mechanism by which control can be exercised. So, and as we said earlier this week, we're very open to including this in this particular section. This leads to a more general comment that we wanted to make about the importance of the compromises that we've reached over the years. And I think, Chair, at the beginning of the week, you said that if, you know, if a delegation does not want, if a delegation wants to retain, it should explain, I think, as an allusion back to previous compromises. And just to be very, very clear, I think our delegation has been very open-minded, and we have challenged all delegations to challenge us Just because we've agreed on something in the past doesn't mean we shouldn't continue to debate, and we're the last delegation that is going to try to convince another delegation not to continue to make their points in exactly the way they want to make them, and we are very open to being convinced that we're wrong. But I do think that, and we're hearing this a lot, and that's fine, I think most delegations are very interested, including our own, in coming back to previously agreed language and trying to innovate. But I do think it's important to recognize that there are two sides to this coin. And the fact that we spend a lot of time working out compromises, I do think means that we should also explain when we don't want to retain. And the compromises we reach are very important, and they reflect foundational compromises. So this is one of these things where it's a balance between the two. But the compromises that we've reached for us really do serve as a starting point in these discussions. We're open to continuing the debate. But we also do recognize that when a delegation is pointed to something that reflects an earlier compromise, that we need to respect that for all delegations, you know, we're all making compromises. We've made compromises throughout this process. So we do want to emphasize that we want to be mindful of the importance and the value of these earlier sort of agreed compromises. And I think with that, Chair, will stop, but very happy to hear other delegations' views and very happy to continue to participate in this discussion this afternoon. Thank you.

Let me thank the distinguished representative of the United States for voicing the views of his delegation and really diving into depth with regard to the questions. Yes, I have said different things at different moments also. I have also said to delegations who want to delete things that just the deletion is not helpful, a proposal to delete is not helpful, because we need an explanation why. So everything, in a way, has to be explained. And I think I read out the mandate three times, four times, and of course, in the mandate, it's also building upon the recommendations and conclusions of the group. It's there. And your position is completely legitimate, of course. Let there be no misunderstandings of that. On the other hand, it was also said by others in the room, we have to now develop elements of an instrument, which we were not doing in the past. So there is this kind of a natural tension, I would say, on what we are doing, because otherwise we could, well, I've said it before, just cut and paste and then think we have elements of an instrument and I don't think that would work either. In your, I think, in your comments, there was of course a very important sentence, I think, that is with regard to autonomous control being the control mentioned in the And, and, uh, the, um, it's good that you remind us of your position on that. You have been very consistent on that also in, in previous, uh, uh, meetings. Uh, I remember when you were sitting at, at the back of the room over there and indeed stating that in the view of the US, there are already systems that can do that. I, I think that, um, so that, that is at least on my, uh, on my radar. Again, thank you for that. Next on my list is the distinguished representative of Austria. You have the floor.

Thank you very much. Thank you for the questions. And after the US chess, we will follow up with some poorly performed Austrian classical music. And I mean, we will try to get through them, and maybe I get stuck because some of the questions are very long, and I will not remember what we have discussed among ourselves how to answer, but just to go to the key questions, maybe at the beginning on human control and judgment, I think when you read the question out, there was also the question, does human control require human judgment or vice versa, was one of the questions, as I understood it, it's not on the screen, but it's something that we also want to answer. I mean, now our understanding control includes judgment, but judgment does not include control. And, um, control is an, an active element, it's something that, that has to be done, while judgment is, is, is something that is executed, but it does not have this, um, active element. We are here more in the semantics, but I think it's something that is also important to be understood. And the other maybe more practical element and the difference between human control and human judgment is that human control indicates that there is a direct link between the human intention, or at some point it was called as a human agency, and the outcome. And as we have stated, and as it was discussed and explained many times, this is linked to requirements under IHL on predictability and other very important elements of international agreements, in particular Additional Protocol I. So there is a qualitative difference and some additional element that we bring in that also relates to the technology that we are discussing here. So we see a qualitative difference between human control and human judgment, and it also serves our purpose better. The judgment has its role. I mean, we do not contest that. It definitely is needed. So we can go along with your proposal here to have both of the elements included. We had a proposal brought forward earlier this week where we would actually have the judgment element included under six in box three, but more related to informed judgments about the anticipated outcomes of the use of force in accordance with legal obligations. I mean, this is some other way to link it and to also explain the difference between control and judgment. One is the active element. The other one is something that It's more the, let's call it, the intellectual element under this relationship between what a human wants to do and what actually happens on the battlefield and how use of force works. Then coming to the next question, which was about at what stage is this context appropriate human control and judgment necessary to comply with IHL. which is a very different question as well. The answer probably lies in the word context appropriate. So depending on what kind of context you're in, the stage could be a earlier one or a later one. I mean, we are going here a little bit beyond our position probably. especially when we don't talk about meaningful human control, but both the words context appropriate, but also meaningful, give an indication that this stage could be at different places. And this is probably the biggest advantage of context appropriate, but also of meaningful. And we should not forget that meaningful is also a qualifier that does not only impose additional things that states have to do, it also gives the whole thing a certain qualification. But if we talk only about the context appropriate, the context defines at what stages human control and judgment should be performed. In the best case, it's as close to the user's force as possible. But this is something that we are not discussing yet, and that is probably also a difficult topic to discuss. Next question, if direct control or oversight in these different stages is always needed to comply with IHL. And the question here is probably a no but. So it's not always needed. And also here, and maybe I'm repeating myself, it depends on the context and on how the human control is performed. And it also, again, has to do with the meaningfulness. The next question, do we acknowledge that in certain systems or situations, human control and judgment can be exercised before the system autonomously identifies, selects, and engages a target? I think in this question, there is, again, a contextual element, and this has to do with time and how. So the question when and how this is done, we have the word "before." The question is when before. immediately before with a close link, or is it done weeks or days or hours before so that there is a significant difference in the environment in which it is -- the human control and judgment should be performed with. So the when and also the how with that kind of extent of human control this is performed makes a difference. And this is maybe the two questions we should focus on. What we are doing in the different boxes at the moment is we are setting up certain ideas and also limitations how this could be done. So by, it does not really relate maybe to this question, but to some other ideas of context and judgment, by setting up the different limitations under 6C, we are trying to create an environment where this makes more sense, where it's easier for humans to perform human control and judgment, and where, and we have to be very honest about this, at some point it will be also creating the possibility for a human to follow the actions that are evolving in front of him. That there is not a race for speed or uncontrollable system where the actions on the battlefield are just slipping out of the hand of the persons responsible for the actions and then also having to burden themselves with the accountability. So many of these questions we are trying to answer in 6C and creating an environment in which context appropriate human control and judgment is even possible. And that's why many of the answers to this question actually lies in the document that we are already discussing. And I think that's also the question to--the answer to the next question. So I will not go into this external factors. These are very difficult questions to answer. I think there we need some practical insight and something that I think is a good practice when you elaborate instruments. I mean, we are just elaborating elements of an instrument, is that when discussions or negotiations get more concrete, that you go into scenarios or practice building. So maybe at some point of our discussions under this mandate or when we go into negotiations, we should maybe also go into scenario exercises where we discuss what kind of scenario, what kind of context are we talking about. We are probably, as it stands now, using a very broad understanding of context-appropriate human control and judgment. But it's probably important that we are also having a similar understanding. And for that similar understanding and also an understanding of different scenarios, a scenario exercise could be extremely helpful. And this is just a suggestion that also was triggered by your questions and maybe some other ideas in that direction come up in the course of this afternoon. Thank you.

Thank you very much for that. Next on my list is the distinguished representative of Ukraine. So you have the floor.

Mr. Chairman, short remark. In my opinion, I'd like to continue the position of Austria regarding that for IHL, Very important is only action, not judgment. In this regard, criminal judgment cannot prosecuted by love. In my opinion, it's not role. What is judgment is human inside brain. Very important is only action, results of action. In this regard, I think that we can consider it human control process only, and inside this control process, judgment is implemented as study, decision making. My proposal is to use only the first part of this text, context-appropriate human control, or only human control process. Thank you.

Let me thank the distinguished representative of Ukraine for sharing his views with the group. Are there any other delegations who would like to reflect on the questions? We have Yes, I recognize Melos. You have the floor.

Thank you very much, Chairman. Sometimes the questions are more important than the answers. That's what Socrates thought, anyway. I think that these issues that you're highlighting for us through these questions, giving us ways to resolve the issues in our document. Now, maybe I'm mistaken, but I think we're very close to the issues that we haven't named, but which are very much on our radar. The issue of international legal responsibility, and I think attitude towards laws must be just as demanding as it is to conventional weapon systems competence and that's a situation of height when there's a situation of heightened danger now to answer the first question about human control and judgment here there's a very important criteria that you have defined if a car If a machine assumes functions of human control, like judgment or whatever it may be, an armed system, if a weapon system, then that then it becomes then it becomes a law as an autonomous system. The system can't be more or less autonomous either way. If the machine assumes those functions, that already makes it a lethal autonomous weapons system. But turning to responsibility, there are a few elements. It seems that the party that uses laws, the country, assumes full responsibility for the use thereof. Why do we use that criteria? It is so as to rule out situations where a machine guided by by a person that its use thereof leads to a violation of IHL. Either way, if such a situation takes place and IHL is violated, the country using laws assumes full responsibility. But if there is present in that chain malicious thoughts, if it's ill-intended at any point when it comes to designing the program, when it comes to decision making, the use of that weapons system, then we're talking about exacerbating that responsibility. Now that might carry criminal responsibility for the persons in question. So it seems we mustn't allow We mustn't create a vacuum when laws is used and where it's considered as being collateral damage. Whatever happens, violations of IHL, even if they took place unintended, they must nevertheless carry international legal responsibility. But in those cases where human control and judgment were present, then that responsibility is heightened. Now that might allow us to get away from sort of relative constructions, relativism, that in that context the control exercised was sufficient. Any replacement by the machine of the human functions of control and judgment at any point, it must place that system, it must basically put it in the category of a laws system. Thank you.

Let me thank the distinguished representative of Belarus for sharing his views with regard to the questions. I see Sweden. Let me pass the floor to the distinguished representative of Sweden.

Thank you, Chair. I guess this is our attempt to answer some of your questions this late in the day. As we mentioned previously, this week, we can support context-appropriate human control and judgment as a compromise. We believe that existing IHL indirectly requires control. We don't see how it would be possible to use a weapon system in line with IHL without controlling it. We further believe that we are here to make that implied requirement explicit. We share the belief raised by other delegations that IHL applies to humans, not machines. Therefore, we think placing human before control makes sense. We do not believe that doing so excludes the possibility that humans use software to control the weapon system. We further believe that IHL requires those who plan or decide upon an attack to make certain judgments in the conduct of hostilities. For us, IHL can be respected and implemented through a number of measures and in various stages of the planning and execution of an attack. Also, not all possible IHL questions arise in all type of attacks. This depends on the context, hence the need for context appropriate. For us, autonomy means that the rules and principles of international law, in particular IHL, should be considered throughout the entire life cycle of laws. IHL compliance can be upheld through a variety of measures, and autonomy may change but never diminish the responsibility of humans to fulfill IHL. This is what we believe that this terminology serves to clarify. Thank you, Chair.

And we thank the distinguished representative of Sweden for his remarks. I think I see only some old plaques. Yeah, I think so. Ukraine, same. Sorry, I couldn't see you. The distinguished representative of Peru, you have the floor.

Thank you, Chairperson. I wanted to comment on some of the questions that you have kindly put forward and specifically regarding the specific contextual factors should be taken into account when deciding on the necessary amount of control, appropriate human control and judgment. I wanted to point out that for my delegation, which is part of the G16, and you know that we have also expressed our preference for another formulation of this concept, but of course, we are now discussing its content, so not the label. We thought it would be important that the notion of context appropriate human control and judgment includes elements like the ability to redefine or modify the weapon systems objectives or missions. or otherwise adapted to the environment to deactivate it, abort, terminate, or interrupt its operation and use as needed, and to constrain its function to self-initiate. This is one. And another aspect or element is the ability to limit the scope and scale of use of the weapon system including temporal and spatial limits and to restrict its targeting parameters and targeting capability. I thank you.

Let me thank the distinguished representative of Peru for responding to the questions. Let me Let's see, is there any other request for the floor? If that's not the case, oh, sorry, I see, I recognize the distinguished representative of the UK. Sir, you have the floor.

Thank you, Mr. Chair. These are very complex questions, and you will note that the UK has provided quite lengthy papers previously on the whole concept of human control and they are available to read on the website. And we will be looking at providing something, a much more comprehensive response, which we can provide to you in due course. But just trying to, in the interests of conversation, just trying to do a couple of immediate responses to the question. So question one, how does human, how do human control and judgment relate to each other? So in the UK, we regard the term human judgment as the ability to judge, make a decision or form an opinion objectively, authoritatively and wisely, especially in matters affecting action. In the case of laws, this will include decisions over where, when and how a law should be used, the risks to compliance with IHL, likely outcomes and what means are needed to control the effects of laws. So the term control, in particular respect to laws, we will see that as being achieved through a range of means, including technological, operational processes and procedures and other precautions across the life cycle of a system. For example, system design, system training, setting of mission parameters, the limiting of autonomous behaviors and operational processes. And there will be more. Question two, at what stages is context appropriate human control and judgment necessary to comply with IHL? So. We used, we utilized the term context appropriate human involvement in our national AI policy. And this concept acknowledges that the appropriate degree of system autonomy and the type of human control or involvement needs to be considered carefully on a case by case basis. The diverse nature of operational environments, context of use, laws, technologies, and capabilities means that no one approach to human judgment, control, involvement, however, whichever term you want to use, will be appropriate to all possible use cases of a lethal autonomous weapon system. UK policy is that appropriate human involvement, control, judgment, again, whichever term you want to use, must be realized at numerous points throughout the system life cycle. utilising authorised, suitably qualified and experienced people exercising judgment to influence, direct or limit the behaviour of an autonomous system and its effects. Those processes will include research and development, testing and evaluation, training and planning for use, article 36 weapon reviews, all the way up to deployment and use. Third question, is direct control or oversight during identification, selection, and engagement stages always needed to comply with IHL? So we believe that you can exert satisfactory and rigorous human control over systems with autonomy without always requiring some form of real-time human supervision. And our approach in that is captured in the various working papers that we have produced, and I won't go into them at length here. The next two questions on contextual factors and external factors, there will be many, and again our papers highlight that, and we stand ready to provide a much more comprehensive response to that to this group. Thank you, Mr. Chair. Let.

Me thank the distinguished representative of the UK for responding to the questions and indeed also referring to the papers which are indeed accessible through the website. Um, okay. Um, I think we are at the end of the discussion with regards to the proposals we made, uh, and, sorry, I see, I see one more request, um, from the distinguished representative of India. You have the floor.

Thank you, Chair. I just wanted to make a few brief remarks on the very interesting and thought-provoking questions that you have on the screen. We've heard the remarks and views of the other delegations in the room as well. And if you just allow me a few minutes, I also want to share our views. It is our belief that the human element is the vital element that we are discussing here to ensure that the weapon systems under discussion could be used in a manner that do not violate the established principles of international humanitarian law. It is this link of human involvement which, in the manner in which the decisions around delegation of taking life and death decisions to a machine, Those are the legal and ethical challenges that are raised, and the human involvement would, in our view, address some of those concerns, and that is the link to IHL. So having laid out that broad principle of the human involvement, we have, over the past three days, in various sections of the text during discussion, shared our assessment that human judgment is a broad, cognitive, broad term that reflects a cognitive function that is unique to a human being, and it allows the human being to assess the situation in a context appropriate manner and make a decision of what kind of control or oversight or intervention is possible. Therefore, through the exercise of your cognitive abilities of a human being, through the exercise of human judgment, we believe that control is automatically being able to be exercised, whether it is through a direct intervention of a human pressing a machine or pressing a button, or it is through any other means, through a software, through other elements of control. So in our view, human and control and judgment are related to each other. But judgment is the broad term under which control could be exercised and not the other way around, which is why it is our request that the text throughout the body of this rolling text reflect the concept of human judgment, which would contain the different ideas of command, control, oversight, intervention that the group has been discussing in various ways. With regards to the question around the stage at which context-appropriate human control and judgment is necessary to comply with IHL, I think, Chair, we would just like to clarify here that in the first box, when we are talking about characterization, we are talking about specifically looking at those weapon systems and characterizing those weapon systems as laws, which do not have any human element or any ability for a human to exercise control or intervene in the execution of three critical tasks around identification, selection and engagement, which is the view of our delegation and other delegations, that those are the three critical functions. So if a human controller intervention is exercised in those three stages, then we are automatically looking at weapon systems that would not be laws, as per the discussion in this group. So I think the context appropriate human control and judgment is not necessarily to be exercised in the execution of those three critical functions, which is the reason we are discussing the whole idea of what constitute laws, what characterizes laws. Context appropriate human judgment could be exercised at stages prior to the activation of those three critical functions, and that would also take care of the element of human judgment and oversight, which would be important to make sure that the principles of IHL are not violated. And with regards you know, what is the, what is the term context appropriate human control and judgment that in certain do we have, uh, do, are we allowing ourselves the flexibility to look at context where that such an human control or judgment could not be exercised? I think again, we are opening up the debate here on at what stage of the life cycle would, the human have to directly intervene. Again, we believe that we must first have some kind of a consensus on what constitute laws, what are the critical functions in which the execution of the function autonomously being done would characterize the weapon as laws. So in summation, I think the point we're trying to make is that we conceptually completely agree with the idea that human oversight, human involvement in some manner is necessary to ensure that these weapon systems are used in compliance with international humanitarian law. And by using the term human judgment, we believe we are taking care of all possible ways that involvement could be exercised, either cognitively or through kinetic action or through any other means. And this broad term would allow the group enough flexibility for the states to then take measures in accordance with their national circumstances in relation to the operational context, the security considerations, domestic legal frameworks to apply these controls in accordance to the specific context. So thank you, Chair.

Let me thank the distinguished representative of India for sharing the views of her delegation with regard to the questions put forward by the Chair. I have a few more. Yes, I think, could we see Mauritania? Is that correct? Yes, you have the floor, sir.

Bonsoir.

Good afternoon. As an observer, speaking obviously as an observer, I would like to, through my statement, enrich the discussion which we've had, as requested by the Chair, by presenting a weapons system which, in my opinion, forms a textbook case so that texts can define as far as possible the most as many cases as possible. This system functions autonomously since there is a complete absence of a command and control chain. But that system has the specificity that it is detonated by the target itself. That system is based on the movement detection mechanism in a given direction and attacks through fragmentation. I would like to, I'd be curious to know the interpretation and judgment under IHL of this kind of system, if it is, would it be considered laws or not? Again, I reiterate that that system is detonated by the target him or herself itself. Thank you.

If I understand you correctly, you could be referring to systems that I think existed until the end of the 80s at certain places in the middle of Germany. But there's not in a way a decision made by the system, it's triggered, like also a landmine is triggered. But I'm not an expert. I'm just trying to squeeze text out of experts. That's my function, I think, as a chair here. I see a few more requests for the floor. The Republic of Korea, Ireland, and Mexico. Let me first pass the floor to the Republic of Korea. You have the floor.

Thank you, Mr. Chair. I might not answer your question directly, but I'd like to discuss at the term the context of appropriate human control and judgment in a way that may answer some of your questions. So first about the presumed importance of human control in IHL. So, while the technological advancements were unforeseen, we find it difficult to agree that human control was always deemed indispensable to comply with IHL. As IHL aims to limit the effects of armed conflict, what matters here is controlling, not necessarily human control. In other words, it's about limiting or controlling, not about controlled by what. And moreover, we need to recognize that the human involvement in the use of force has long been diminished, and autonomy is a spectrum, and it's being used today. We are not living in a time of sword and shield, but fighter jets and long-range missiles. So the emphasis on human control at every point of the life cycle makes less sense. Secondly, about human judgment. The human judgment is essential because it is inherently context-dependent, as it is described by our distinguished delegation from UK and India, and the very act of judgment ensures it is context appropriate. So therefore, we believe the human judgment alone may be sufficient to ensure laws can comply with IHL. Thank you.

Thank the distinguished representative of the Republic of Korea for reacting to the questions and also to the previous remarks by colleagues. Next on my list is Ireland, to be followed by Mexico. But let me now pass the floor to the distinguished representative of Ireland. You have the floor.

Thank you, Mr. Chair. And while we're still considering your guiding questions, we just wanted to engage a bit in the interactive exchange. So first, we want to echo the view of our Swedish colleagues that IHL applies to humans and not machines. And we agree with our Austrian colleagues that control necessitates judgment and is the direct link between human agency and outcome, which is crucial for IHL compliance and accountability. We think meaningful could be a better construction as previously expressed, but we do understand that the current construction reflects a compromise. We just wanted to accept the invitation from our U.S. colleagues by contributing to one point that they made regarding weapons that exist with no human control. And we think the weapons that they might be referring to actually enjoy a high degree of human control, and some might say meaningful amount of human control. They have strict geospatial and temporal limits, as included in box three, that allows for humans to exercise meaningful human control and judgment. We would not argue that those systems have no human control. Indeed, they are not prohibited due to the high degree of context appropriate human control and judgment. And we're happy to discuss that more with our colleagues, regardless of the background music playing, in case we misunderstood. So for us, the existence of these weapon systems does not throw out the necessity of human control. They might actually provide an example of sufficient or meaningful human control falling under regulation of the two-tier approach. And so for us, the formulation in paragraph 5 seems appropriate in this box and elsewhere throughout the text. Thank you, Chair.

Let me thank the distinguished representative of Ireland for sharing his views and commenting on previous remarks. Next on my list is Mexico. You have the floor, sir.

Thank you so much, Chair, for presenting these questions, and I think I will follow the U.S. in supposing we're not in Geneva, but in Montreux in the jazz festival. And I'll be riffing here, obviously, without being a definite position with regard to these questions. So I think, first, with regard to the first question, I think the starting point that it's important is that there has been a recognition expressed in-- common recognition expressed in this GGE that uncontrolled weapons systems would be both strategically inconvenient and militarily ineffective. Moreover, there has been an understanding that while speaking of laws, we're recognizing that while existing IHL rules are fundamental, we are dealing with an innovative category of weapons that We all, at the issue, agree that they present new challenges, and that's why we created this GGE, as they raise fundamental question as how those rules apply and if there should be new rules needed or particular measures related to these specific weapons. So first, it is necessary for a human to make the judgment that matches legal evaluations and ethical considerations with an expected use of force. This judgment is done both in the abstract, for example, while determining if the means of methods of warfare would be in all or some circumstances contrary to IHL by its nature of the weapon, or by its use that needs to evaluate in particular context the legality of its use. However, as others have said, while judgment is evaluation of the weapon itself or with regard to a context, with an appropriate situational awareness, there is the need for the human to have control in the operational aspect. And this has two dimensions, the control of the weapon itself, which goes from its understanding, the decision on when it should be used, and in certain situations, to be able to deactivate or terminate the attack, and also being able to intervene when that situation has changed. And it should also be able to influence the system when there are biases or unexpected algorithmic evaluations with regard to the context, the intention, or the character of the target. Moreover, the context will be important to be determined both with regard to an understanding that not all weapons can or should be used in any situation or circumstance. And it needs to also do a matching of the weapons characteristics, both of its functioning on the level of kinetic power or action that they produce, and the situational characteristics. So what do they represent in practice? I think it would be matched to the particular prohibitions and regulations that we are considering in box three throughout the life cycle. It won't be a one answer to each single-- a single answer to every situation, but it would be dependent on the context. But while the judgment or control could vary depending on such context, we could not envision its absence. So I think that's how, from this rethinking, I answer this question.

I would like to thank the distinguished representative of Mexico for sharing his views with regard to the questions put on the screen. I see Austria in front of us. If you don't mind, I would like to take France first because it's their first reaction to, I saw they were working quite hard to get something ready. So please, you have the floor.

Thank you very much. I'll be very brief so Austria can take the floor. Now, to try to answer your first question, according to France, it can be deduced from international humanitarian law that we need a certain form of human control and judgment. The notion of human control covers both physical, operational control over the system as The judgment is related to different target identification, appreciations, and legal qualifications. And in order to add a point related to your last question, the degree of control will be appreciated and determined by the commanding officer based on external factors. Thank you.

Thank you very much, um, to the distinguished representative of France for her replies to the questions. Um, I have, uh, Austria. You have the floor, sir.

And this is just a, a coda to my previous performance to make some corrections or else I add some arpeggios or other things. We, we realized, uh, listening to other colleagues and the very good interventions we heard from, uh, Ireland, Mexico, and now from France, um, that.

We, uh, were a little, little.

Bit led astray by your questions into talking about human control as something that is, uh, a kind of a single measure, because there is the word exercise in, in your fourth question, um, how is human control and judgment can be exercised. And the question, and it was also answered now that it has to be a, a, a, um, a continuous thing, and human control, and this is something, and here we are, uh, here we are having our previous working papers where we are talking about human control is something that has to be achieved through measures, and this is also our approach to box three, so it is something that, um, has to be achieved and human control is necessary to then achieve the higher goal or the normative goal of complying with IHL. So human control is linked to the IHL level, so it's something that has to be there constant and the question is not when and how it has to be exercised, but through which measures, in which context, when and how.

It has to be exercised.

So in your questions, we should be rather talking about the measures, but not the human control. Human control should not be vary. It should be something permanent that should be achieved. And in many cases, and here we agree with Ireland, in many cases, in many weapon systems, it is already there to some extent through the nature of the system, especially when it comes to many of the existing ones. And what we are trying to do here is to spell this out and to make this kind of categories are more the different limitations that would apply to that and make it possible to achieve this human control. Just to have an explanation and correct our mistake from the previous answers to your questions. Thank you.

Let me thank the distinguished representative of Austria for taking the floor. Oh, sorry, I was hidden behind the Um, I'd now pass the floor to the distinguished representative of Pakistan. You have the floor.

Thank you, Chair. We would also like to very briefly comment on some of the discussions that have taken place, uh, specifically some of the points raised by delegations of the U.S., Ireland, and now Austria. about the existing examples in certain weapon systems. It has been pointed out that there are weapon systems out there which are capable of autonomously selecting and engaging targets without direct human involvement at precise moment of engagement. And we believe that, yes, even in those cases, it does not imply an absence of human control, but it demonstrates precisely that the context-specific nature of human control that is required. For instance, if we consider the air defense systems or the close-in weapon systems, the crucial parameters, as the delegation of Ireland also highlighted, are embedded deeply within the operational framework of those weapon systems. So although those systems may autonomously perform certain functions and engage incoming threats Human retains control by presetting the operational parameters, including those geographical limitations, temporal limitations as well, to ensure that the systems remain predictable, accountable, and compliant with the IHL. Thus, we believe that the control need not always be a real-time intervention at the exact moment of engagement, and this was just pointed out by Austria as well, perhaps in response to one of the questions that at what moment it needs to be exercised. In our view, it can and it frequently does occur through robust human involvement at earlier stages also. And thus, we use this concept of context appropriate human control. This is exactly why we believe that this broad formulation that you have proposed is essential. It accommodates both the existing practices and also takes into account the future technologies that we are trying to address. Thank you.

Let me thank the distinguished representative of Pakistan. And I now have a request for the floor from the distinguished representative of Brazil. So you have the floor.

I think this debate is extremely interesting and I was listening actively to it and trying to think of ways where we could make a contribution to this jazz jam session that we are engaged right now, especially since jazz is one of my favorite genres, even though Sometimes it feels like we need to go with the flow in order to actually get to something. And the flow has been given to me by my colleague from Pakistan, which made some of the points that I wanted to make. What I can contribute to this, I think, is that we need to dispel a notion that sometimes that periodically recurs in our debates about this. that human control means direct control by humans of laws at all times. I don't think that this is a notion that anyone here espouses. Just as autonomy is not a binary concept. A weapon is either autonomous or not. It exists on a spectrum. Human control should also be understood as a spectrum, and a spectrum that is determined by the context in which it is exercised and over what it is exercised. I think it's easier, just like my Pakistani colleague said earlier, to understand this by resorting to systems that are already operational and which are well understood and everyone know more or less how they operate, which are air defense systems, and close-in weapon systems. I think we all agree these are autonomous weapons, or at least weapons capable of autonomy. And we all agree that they can be operated under judgment and control. We don't just exert judgment and then activate these systems and hope for the best. They're always under control. A close-in weapon system can be operated safely at sea in not in close proximity to, or not when in combat operations and not in close proximity to civilian vessels. But if the same system operated in the very same way is activated in peacetime in a port city, I think we all agree this would not be a situation in which this weapon is under control. 'Cause even a pleasure boat ride going in the direction of the system can be interpreted to the direction of the vessel can be interpreted by the system as a threat and therefore be engaged. So I think this is a very vivid and easily understood way in which both autonomy exists and how control over autonomy can and is already implemented. It's not the human control at all times, but human control over the effects of the weapons that we are talking about here. And I hope that our debates about box three, in which this notion of human control and judgment will be discussed at more detail, will help us reach a common understanding on this subject. Thank you.

And let me thank the distinguished representative of Brazil for his comments. And indeed, in many ways, we have all While discussing the questions, we also, in a way, have provided some background to the possible exclusion clause, indeed, that certain systems do exist already. They have autonomy, but most likely there's no need to have further regulations on them. I see one more request for the floor from the distinguished representative of Italy. You have the floor, ma'am.

Thank you, Chair, for presenting to us questions that may help framing the discussions on this relevant section. These questions, as mentioned by the distinguished UK delegation, are complex and would require more time to provide comprehensive answers. We will still try to provide our perspective on a few points. In terms of how human control and judgment relate to each other, for us, exercising judgment is an action which entails a decision. We consider the systems operating under a structured military chain of command and control and in a specific context in which the human, after assessing the situation and assuming responsibility, which for us means exercising judgment, critical reasoning, considering many factors such as operational context, assessment, characteristics, and capabilities of the weapon, will assign the execution of specific tasks related to critical functions within a clearly defined operational framework. After exercising judgment with regard to compliance with rules and principles of IHL, namely distinction, proportionality, and precaution in attack, the human operator takes in consideration use of force. In all the phases of assessment, there will be a continuous human involvement before taking the decision of using LoWS. A distinguished delegation mentioned the other day that we do not need labels, but shared understandings, nevertheless, for us, this kind of involvement is a control that could be exercised even before the three critical functions of laws, and I thank you, Chair.

I see another request for the floor from the distinguished representative of Israel. You have the floor, sir.

Thank you so much, Chair. And we have been following this interesting exchange of ideas, so we wanted to share some thoughts of our own. And before that, just in the spirit of consensus, I would like to quote the musician Frank Zappa, who said, "Truth is not beauty, beauty is not love, love is not music, music is the best." And now for our comments. Chair, following the questions that were raised and the questions -- sorry, in the discussion that developed here, a few general points we would like to make. Firstly, as we said before, it is critical to distinguish between primary rules of IHL on the one hand, including categorical prohibitions on weapons, targeting rules, and other legal rules, and on the other hand, legal or practical aspects which are not themselves primary IHL rules. This includes the issue of human-machine interaction. While this concept might have a practical meaning in the implementation of IHL, treating it as IHL rule or using it to formulate new prohibitions will be a mistake on a number of legal and practical levels. Secondly, we believe that the appropriate type and level of human-machine interaction will be determined by the characteristics and capabilities of the particular system in question vis-a-vis the operational circumstances and the mission's requirements. In other words, there is no one fits all set of requirements that every weapon system needs to meet, but rather these requirements may be adapted appropriately. Accordingly, temporal and spatial restrictions may be applied to the operation of the system during its development, deployment, or before specific uses, depending on the case. as necessary in order to ensure lawful use in the intended circumstances for use. For example, it is possible that an operation with a LOAS will be limited to a military base inhabited solely by combatants in which the weapon system will operate for a limited frame time if this is necessary to achieve compliance with IHL. In this context, Israel believes that human judgment will always be an integral part of any process regarding laws and will be applied during the life cycle. Weapon systems based on emerging technologies in the area of laws will operate as designed and programmed by humans. Thank you, Chair.

Let me thank the distinguished representative of Israel for sharing his views and also commenting on the views from other delegations. We started with jazz, we ended with Frank Zappa, and I don't want to make an opera out of it, especially not a Wagnerian opera, because then we would be here for another few hours, and we will lose the interpreters. really like to thank you from the bottom of my heart for, for your, uh, for your taking part in this discussion and indeed for, for reacting to the questions, um, uh, yes, as the colleague from Belarus said, sometimes, um, the question is more important than the answer, but, um, I, I also very much appreciate the answers you have given, uh, clear, we still have different views, but we at least have a different, we have a better insight in why we have differences of opinion with regard to, um, human control and judgment, um, and whether it should be context appropriate or not, uh, or control and all the other variants that we have, uh, discussed, uh, this afternoon. That's certainly helpful and as was stated also by others, this is not the only place in the text where we deal with that, it comes back in the next box, so this has been also an excellent preparation for the discussion on Box 3. Um, while we were discussing box two here, uh, my team has been very diligently working on, um, some amendments to the text of box three, um, we will do a last review, uh, in the coming half an hour or so, and then we will mail that, um, to your, um, to your delegations, so we will be able to continue our discussion then tomorrow morning, uh, at, uh, at 10 o'clock, um, and we have then, uh, a slightly amended, uh, text, uh, of box three in front of us, uh, and we can discuss that, and you have some time also to prepare yourself for the discussion we have tomorrow morning. Um, let me again thank you for your participation and for your very constructive attitude. Um, this meeting stands adjourned.