6th Meeting - 2nd Session Group of Governmental Experts on Lethal Autonomous Weapons Systems 2025

Excellencies, distinguished guests, could I ask you to take your seats again? We still have two boxes to go. And you might want to share some observations about those two boxes. If you don't, it's fine with me, but then I have certain conclusions ready. But I think we have some work to do. Good. Let me ask the Secretary to display box four of the rolling text on the screen. There we have it. In this box, among other things, I try to clarify the measure contained in power one by explicitly referring to the obligation to conduct so-called legal reviews in accordance with states' obligations under international law. Furthermore, Following suggestions made on the different risks regarding bias during the first session in March, I clarified the distinction between, on the one hand, unintended bias in AI capabilities in power six, for instance, related to any bias in the data sets used, and on the other hand, the risk of automation bias in power seven related to the humans operating the system. I also added new language in para 8 to ensure appropriate guidance, training and instructions, highlighting that this is mainly aimed at those in a responsible chain of command and control, or somebody would say in a responsible chain of command or exercising command and control, which includes direct users of laws. At the end of the box, I also added a new para, para 9, with a clear link to compliance with the IHL and the different stages in the life cycle of the laws, aiming to cover also situations beyond the direct development and deployment of laws that cannot be used in line with IHL. So far from my side on box four. So let me now open the floor for delegations to share their views with the room. We have two plaques identified, the Republic of Korea, but also one at the back we were not able to see clearly, so if the person could... Is it advocacy for human rights? Is this advocacy for human rights? Campaign to stop killer robots. What is it? Oh, campaign to stop killer robots. Okay, that's the number. Sorry, we forgot to bring our... telescopes because I have glasses, but that goes -- this is not covered by the insurance, I think. And we have one more. Check, yeah. Okay. So let me start by giving the floor to the Republic of Korea. You have the floor.

Thank you, Mr. Chair, and good afternoon to all. And I will break the ice for the box four. But before turning to box 4, my delegation would like to briefly comment on box 3 in response to the suggestions made by different delegations. First, in our previous statement, we proposed removing paragraph 6C altogether, as it is unnecessary to restrict the flexibility of laws. That said, we can support the French proposal to insert the word "significantly" before "modified." And with respect to paragraph 6D, Roman numeral 4, Having considered the different opinions expressed, we are prepared to support the changes proposed by Singapore and many other delegations to delete the phrase "by nature". At the same time, we reiterate our position that paragraph 6 in this box should be placed under box 4. Now turning to box 4, we believe this section is already in good shape, but we would like to make three suggestions. First, with regard to the chapeau of box 4, And as we have underlined previously, it is important to recall our shared understanding that our works do not impede technological progress for peaceful purposes as emerging technology presents both opportunities and challenges. To that end, we suggest incorporating the agreed language from Guiding Principle J of 2019 so that the chapeau would read, "Without hampering progress in or access to peaceful uses of intelligent autonomy, technologies prior to potential deployment and as appropriate throughout the entire life cycle of laws, states should. And we believe that spelling out this broad principle at the outset will help frame the general direction of this box. Second, given the unique characteristic of laws which are likely to undergo more frequent and incremental changes compared to traditional weapon systems, we find it is overly restrictive to require legal reviews for every instance of further development or modification. So to address this, we suggest inserting the word "significant" before "impact," thereby making the provision more practical. I'm talking about the para 4 here, inserting "significant" before "impact." Lastly, as my delegation pointed out during the informal consultations and with, along with other delegations, that paragraph 9, referencing to limiting the manufacture, remains problematic to my delegation. We understand the manufacture here to mean mass production. not small scale production for testing and research. In this connection, and coming back to my first point, it is important to ensure that those measures do not necessarily hinder technological advancement. So we do not have any specific texture yet for this part, but would welcome any revisions that can address this concern. Thank you, Mr. Chair.

I thank the distinguished representative of the Republic of Korea for sharing the views of the delegation and making some concrete comments and suggestions. Next on my list is the Campaign to Stop Killer Robots, to be followed by Czechia and Mexico. But first, the distinguished representative of the Campaign to Stop Killer Robots, you have the floor.

Thank you, Chair. One key element of meaningful human control over weapon systems, alongside the restriction of the parameters of operation of systems, that States have begun to address in box 3, is the need for users to have a sufficient understanding of systems and how they will interact with the context of use. We therefore welcome the elaboration in box four of some of the elements and requirements that can contribute to such understanding. Again, we urge States to move to negotiations on a treaty, establish in a legal instrument that the functional understanding of systems is a core requirement, and elaborate the elements of this. We welcome the separation in the current draft of the very different types of bias that have been under discussion by states. We read paragraph six on unintended bias to include autonomous systems reproducing in their effects structures of bias and discrimination that exist in our societies, such as sexism, racism, and ableism. For stop killer robots, when it comes to life and death decisions, deciding who to kill, the reproduction of biases that will cause disproportionate harm to already marginalized people is not something that we should merely seek to mitigate. It is abhorrent and fundamentally unacceptable. This is a key reason why we must categorically prohibit systems that target people within a broader legal instrument on autonomous weapons systems. We must draw this red line that this is not something we can countenance. Racial, gender, and other bias is a problem for anti-personnel autonomous weapons systems. much more than it is a problem for systems designed or used to target military material. For this reason, we urge States again to consider the anti-personnel autonomous weapons systems issue more fully and to support a prohibition. Thank you, Chair.

Let me thank the distinguished representative of Kyrgyzstan. Next on my list is Czechia. Let me pass the floor to the distinguished representative of Czechia. Madam, you have the floor.

Thank you, Mr. Chair. As this is the first time that you get the explore, let us first thank you for your chairmanship and your work done on the rolling text, and express our appreciation for the progress made by the group, especially in terms of identifying measures necessary to ensure compliance with IHL. Regarding the box 4, in general, in line with our recently circulated working paper, We would like to underline that risk assessment and mitigation measures must be integrated throughout the entire lifecycle of weapon systems, from design and development to deployment and use, with humans retaining ultimate control and responsibility in relation to the use of force in armed conflict. We especially appreciate the current wording of the first paragraph as we consider legal reviews as required under Article 36 of Additional Protocol I. to be critical in determining whether a new weapon system complies with international law. We also strongly encourage the exchange of national practices and guidelines to support effective implementation of legal reviews. As for paragraphs 5 and 8, however, and for the whole text in that sense, in our view, we should avoid using the term "human user" and use "operator" instead, as "operator" reflects the chain of command under the IHL, rather than the generic "user." We maintain that clarity is key in reaching a tangible result of our efforts. To that end, among others, and also in our working paper, we have endorsed the draft articles on autonomous weapons circulated today and jointly submitted by Australia, Canada, Czechia, Estonia, Japan, Latvia, Lithuania, Poland, the Republic of Korea, the United Kingdom, and the United States. In our view, the draft articles offer a pragmatic and balanced approach and framework and could serve as a constructive basis for our further work.

Thank you, Mr. Chair.

I thank the distinguished representative of Czechia. And indeed, the working paper that was mentioned is also on the website as working paper number five. So for all of you interested, you can find it there. Next on my list is Mexico, to be followed by Australia. But let me first pass the floor to the distinguished representative of Mexico. You have the floor, ma'am.

Thank you very much, Chair. First and foremost, I would like to once again highlight our gratitude to you. We very much appreciate the way in which this box has been drafted and prepared. So thanks to all those who put it together. The overall way in that we're working with these different numbered boxes is a great way of ensuring that we can work strategically and smartly towards a common understanding. Now, in 4.1, we welcome the requirement for legal reviews to be conducted in accordance with the obligations of States under international law. This is an element that my delegation has repeatedly suggested and requested be included, so we're glad to see it there. The only element that we think that could be added to strengthen this paragraph as part of the set of recommendations, to render them more clear, it would be that these legal reviews be periodic, not one-off. We have some language that captures the fact that there should be periodic legal reviews to make sure that they keep pace with technological developments, et cetera. We would like to weave in somewhere the element in paragraph 1 that they need to be periodic legal reviews, if we could do that. I would also like to make a comment picking up on a recommendation made by the International Committee of the Red Cross, the ICRC, with regard to the chapeau and paragraph 1 of this box pertaining to the word "deployment". Although we understand the sense, the meaning that has led to the choice of this word "deployment" in this box, in the chapeau, et cetera, I think perhaps it needs to be replaced or complemented by the use of the word employment. So either employment or deployment and employment. We're not talking just about the deployment, we're talking about its use as in its employment. The thrust of all of this is much larger than this. We need to, about the different types of behavior that we need to be keeping an eye on. It's not just the deployment, but also the employment and or use of these weapons. So I think that if you could consider that suggestion of either replacing deployment with the word employment or using the language deployment and employment, we'd welcome that. Moving now to paragraph 6 and 7 of box 4, we welcome the explicit reference to the need to implement measures to detect, correct or mitigate bias. In that understanding, we think that the current wording is good, but perhaps to capture some of the nuance on the point that the group has worked on in more depth, I don't know whether "detect, correct or mitigate" would be the appropriate wording. It seems a little bit exclusive, so we'd like to explore the possibility of using the wording "detect, correct and/or mitigate". i.e. to make it abundantly clear that you might have to carry out all three activities, you might have to detect and correct and mitigate, rather than giving the idea that you would choose one or the other of the options of the three. We would also welcome some language that contains more specificity about the biases that we're concerned about. One of the concerns that my delegation has is that we make sure that we capture concerns about bias related to discrimination on the basis of race or gender, or of mis- or abusive interpretation of intention or context, et cetera. I think that it might be worth explicitly setting out the forms of bias that we are concerned about. I pick up on another point made by the ICRC. Perhaps The characterisation of unintended bias is a very diplomatic and eloquent way of capturing that concern about bias. We can appreciate the word unintended, but I think that we can beef up our concerns about bias and that we have specific concerns about specific forms of bias, borrowing the language from the Convention itself, for example, harmful bias. What we're trying to avoid is harmful bias, so we would suggest adding the word "harmful bias" in this section, which would give more weight to the need to identify those biases, and adding that word "harmful" would be in line with the language employed in the Convention on Conventional Weapons itself. Thank you very much.

I thank the distinguished representative of Mexico for sharing the views of her delegation and for making concrete proposals. Next on my list is Australia, to be followed by Ukraine, but let me first pass the floor to the distinguished representative of Australia. You have the floor.

Chair, thank you for giving me the floor. Australia supports paragraphs 1 to 5 and 8 and 9. Australia is especially supportive of paragraphs 1 and 4, as this GGE has, over many years, recognised the valuable role legal reviews play in ensuring laws are used in compliance with international law. We would like to focus our comments on paragraphs 6 and 7. Overall, we welcome the amendments made over the intercessional period, especially the inclusion of: unintended bias. We propose to further strengthen paragraphs 6 and 7 by adding the word "and" so that it reads, "Implement measures to detect and correct or mitigate." This would make clear the need for states to have measures of detection, but also to subsequently correct or mitigate the bias found upon detection. Chair, Australia prefers that this GGE continue to have a technology neutral approach, which is important to future proof the text. For this reason, we would propose removing the reference to artificial intelligence capabilities in paragraph six. Artificial intelligence capabilities are just one type of capability that may underpin a laws. A LARS may not necessarily use artificial intelligence. We are concerned that by limiting paragraph 6 to artificial intelligence capabilities, we are implying that only unintended bias in LARS with AI capabilities, but not all LARS, should be addressed. Where more detail is desired, we should refer only to processes or technology which would be guaranteed to be in any LARS. As such, the reference to data sets is appropriate, but would propose replacing AI models with system learning. This is the same language my delegation put forward this morning with respect to box three, paragraph six, capital D, Roman numeral five, and we see it as a more technology neutral approach. An effective and technology neutral paragraph six would therefore read, Implement measures to detect and correct or mitigate as much as possible unintended bias, including bias in data sets and system learning. I thank you.

I thank the distinguished representative of Australia. Next on my list is Ukraine, to be followed by the United Arab Emirates and Japan. But first, the distinguished representative of Ukraine.

Thank you for giving my delegation the floor. The delegation of Ukraine fully welcomes the overall idea of box four. as it reflects several duties of utmost importance placed by international humanitarian law on all states. Firstly, this is the duty to conduct a legal review of new weapons, which can be found in Article 36 of Additional Protocol I, which is of outstanding importance, as has been underlined by the distinguished delegate of Australia. Further, this box addresses the duty to take precautions, notably summarized in Article 57 of Additional Protocol I, and duty to disseminate knowledge about international humanitarian law summarized in Article 83 of the same protocol. Here we cite the protocol as a useful guide while still acknowledging the importance of other instruments where the same or similar obligations can be found, and at the same time in full understanding that not all states found it possible to join Protocol I. However, we believe that all states share these commitments. As for paragraph 1, We're more than supportive of it. Our only concern is that it is highly similar to paragraph five from box three. Our view is that these two paragraphs must be once again reconsidered in order to either merge them or clearly distinguish between them, making it entirely clear that they concern one obligation or different obligations. As for paragraph 2, the delegation of Ukraine upholds the importance of this provision, however, finds it repetitive of paragraph 5 of box 2. We believe that this norm must be a part of box 2, and therefore we see no need in repeating it in box 4. Paragraph 2 of box 4 should therefore be deleted. Speaking of paragraph 3, our delegation would like to seek greater clarification of the obligations contained therein. We believe that it should be very precisely specified, understood by whom, the end user, the human operator, the commander, the manufacturer. Perhaps here we can use the term the chain of command to make sure that we know for certain who is expected to understand the capabilities and limitations of the lethal autonomous weapon systems. Coming to paragraph four. We find it reflective of the obligation contained in Article 36 of Additional Protocol I. Ukraine fully upholds this obligation and our delegation welcomes the specification in more concrete terms that the duty to conduct a legal review of weapons must cover not only new weapons but also their modifications. What we would like to clarify is that such a duty should not be unreasonably demanded, should not unreasonably demand to cover all modifications. The circumstances of the actual battlefield result in numerous petty improvements made on the ground, and not all of them actually affect the weapon capabilities, and not all of them warrant a legal review. We deem it necessary to specify that such impact must not have just effect, but that effect must be significant, a point which, to our opinion, has been on another occasion addressed by the representative of the ICRC, if we understand the speech of the representative of the ICRC correctly, and please correct me if I'm wrong. We understand that such wording is somehow subjective, but we still believe that we must be clear on the point that not all modifications warrant additional review, but only significant ones. What is significant should be decided on the case-by-case basis and on the basis of general legal principles. Paragraph 5, to the opinion of our delegation, continues the pattern of describing the elements of a legal review of weapons. We are hesitant as to its necessity, since it appears to describe not a legal duty to conduct a legal review, but the concrete methods for such review. We are not sure that all such methods can be described in an international instrument. Furthermore, such description has a potential of unduly limiting the range of options available. If we create a list, most probably any entities responsible for the review inside the states will limit themselves to those options exclusively. And that's not a result that we hope to receive. We therefore believe that an international instrument must contain the duty in general, while its contents should be explained in manuals collections of best practices, and other similar instruments like, for example, reports of these group of governmental experts, where such measures have already been mentioned. Paragraphs 6 and 7 are, to the opinion of our delegation, of paramount importance. Eliminating biases is an exercise that can never be entirely completed, but which surely must be rigorously undertaken. Removal of bias is essential for ensuring the correctness of targeting process. To our opinion, elimination of biases is a basic precaution that must take place in the use of lethal autonomous weapons systems. Therefore, we deem it necessary to describe in these paragraphs, to describe these paragraphs using a more general wording applicable to precautions in attack. We further argue that another paragraph, or maybe even several paragraphs, can be added on specific precautions concerning the laws. Such precautions may include, at the very least, measures directed against cyber attacks and data poisoning, possibly also those against jamming and spoofing, although we do not insist on the latter two. As long as the laws is a specific type of a weapon, precautions must also be specific. Paragraph 8 is of special importance for the delegation of Ukraine, a country investing incalculable time and effort into teaching its military personnel the norms of international humanitarian law. Even in the context of an armed conflict, Ukraine does its best to retain humanity and remain within the borders of humanity. We believe that this is best achieved through teaching and learning. Therefore, we support reinforcing of the wording of this paragraph by replacing the words "States should ensure" with the words "States must ensure" to underline the imperative nature of this provision. Our delegation has no specific commentaries on paragraph 9, which to our opinion summarizes the general duty. We could probably find it appropriate to move this paragraph to box 1, describing general obligations. And this is it for Ukraine. Thank you, Chair. Thank you, distinguished delegates.

I thank the distinguished delegate from Ukraine for stating the position of the delegation, making proposals. Next on my list is the United Arab Emirates, to be followed by Japan. Let me now pass the floor to the distinguished representative of the United Arab Emirates.

Madam, you have the floor.

Thank you, Chair. I would like to provide some comments on box 4 of the rolling text. Beginning with paragraph 1, we welcome the reference therein to legal reviews. under article 36 of the first additional protocol of the Geneva Convention, which calls upon states to ensure compliance of any new weapon with international humanitarian law. We are also in favour of sharing best practices in this field, as long as national sovereignty is respected. and that sensitive and confidential information is protected. That's information related to national security and to intellectual property rights. We have always called for the bolstering of the transparency in legal reviews as a way of building confidence. And that is why we support the statement made by certain of those who took the floor before us with regard to legal reviews and the need for transparent cooperation. factoring in national security interests and the need to protect sensitive or exclusive proprietary information. With regard to paragraph 2, we support this paragraph, echoing our previous stance with regard to this paragraph. We have always been of the view that context-appropriate human judgment is important not just at the use of the weapon but throughout the life cycle of the delivery system. or systems as a whole in order to guarantee that the legal dimensions and moral dimensions are duly taken into account. The use of force should never be entrusted solely to these systems absent any human control. There needs to be effective human control. But that effective human control needs to be something that's not just theoretical, but something that's actually implemented and done in practice. Now, our position is therefore compatible with that of those states who called for us to need to plug the gap between theory and practice by specifying the degree of necessary control and oversight. And we echo those states who have insisted that the function of selection of targets must be carried out under human control, as must engagement. Now, on paragraph 3, in our view, this paragraph solves to bolster the dimension of responsibility and accountability in the use of these systems. assessments need to be updated and based on multiple scenarios. We support the inclusion of the element of predictability in evaluations, and we hope that that will become an integral part of the overall approach. Along the same lines, we support those international reports that clarified that the understanding of the different scenarios and capacities and a better understanding thereof will allow for an appropriate interpretation and understanding of the different mechanisms that come into play in this field. We need to identify solutions when it comes to responsible predictability, which is something that we're already undertaking via our strategic partnerships. with regard to predictability, detection, as well as the analysis of the effect or fallout of the use of these systems, which is an important point of the overall system of updating and fine-tuning these systems. Now, on paragraph 4, we agree with the importance of reviewing modifications. Core modifications have an impact on the development of programs when it comes to the selection and engagement. So these evaluations or assessments need to be an integral part of national mechanisms created under national relevant military legislation or law. Modification of programs should not be carried out without legal approval, clear and explicit legal approval. We personally, as a delegation, are of the view that making sure that that is required would serve to boost confidence in this system or in this framework. Now, on 5, here we support the wording in paragraph 5, which calls upon us to implement regional and international measures which will duly factor in the geographical and cultural context. We personally believe that tests that duly factor in the moral dimension are also important. It is also imperative that we establish international cooperation to such a degree that it will serve to boost confidence between states. We echo those states who previously called for a reference to the importance of taking into consideration complex environments and circumstances whenever such systems are deployed. Now, in paragraph 7, we believe that paragraph is compatible with our own position. vis-a-vis automation bias in artificial intelligence or AI, particularly in the military domain. We call for greater transparency in the implementation of training modules or training models rather for AI. And we call for the establishment of mechanisms to root out bias. These need to carry out periodic reviews of bias in AI models. We are strong supporters of bolstering the moral dimension of AI and its use. And that's why we call upon states to roll up their sleeves and begin the hard work of tackling and eradicating biases inherent in artificial intelligence systems. We believe that that's important. because doing so would strengthen the legitimacy of the use of these systems. Turning now to paragraph 8, we would like to underscore the importance of integrating the training referred to in paragraph 8 in military training courses and military courses in order to ensure that it covers all the legal, moral and other dimensions related to this field. I think it would also be important to include a reference to certification, the need for certification, proving the competence of those who are using those systems. So they should be duly certified in their use. In the United Arab Emirates, we support the international or we support an international mechanism or the creation thereof that would be focused on this issue. We fully endorse paragraph 9 because it is compatible with our long-standing position to which combating the unregulated use of this material. And we call for the conducting of a study as a precursor to the establishment of an international mechanism to ensure compliance with IHL. To sum up, we are of the view that box four anchors any regulatory framework of this issue. It's the bedrock of our overall approach. And so it's important that the engagements contained therein are also captured in national legislation, in line with the principles that we have adopted as the USC to wit ensuring human control and oversight, compliance with IHL. and ensuring responsible innovation and development of AI. Those have been our watchwords, our guiding principles, and we believe that that should be true of all partners. We call for such optimal practicals at the international level. We call for partners to provide capacity and support for developing countries so that they can also bolster their capacities in this area. Thank you.

I thank the distinguished representative of the United Arab Emirates. Let me now pass the floor to the distinguished representative of Japan, to be followed by Canada. You have the floor.

Thank you, Mr. Chair. I would like to begin by expressing our appreciation to the working paper submitted by Czechia and react to some of the comments made on box 2. First, Japan supports the proposal made by the distinguished delegation of France to add significant to sub-paragraph 6C. We also support the proposal made by the distinguished delegation of the Republic of Korea and supported by other delegations to move paragraph 6 to box 4. Now moving on to box 4, first on paragraph 4, we consider that laws could include systems that necessitate follow-up after legal review and consider that the text could merit improvement on two points. First, we agree with the proposal by the Republic of Korea and Ukraine to add "significant". as minor technological updates would not require such a review. And secondly, it would be useful to differentiate the nature of legal review in paragraph one, which is a legal requirement, to the review referred in paragraph four. With these points in mind, we would like to propose the following formulation. Conduct appropriate reviews to ensure that modifications that have a significant impact on the target selection and engagement functions or on its anticipated effects are in compliance with IHL. Next on paragraph 9, while we understand the intention of this paragraph to address the whole life cycle of a weapon system, We should also be mindful of the dual use nature of technology and not to hamper technological advancement. Therefore, we propose to modify after laws which are incapable of being used in compliance with IHL under any circumstances. And finally, along the same line, we can consider favorably to the proposal by the distinguished delegation of the Republic of Korea on a possible chapeau. Thank you.

I thank the distinguished representative of Japan for sharing the views of her delegation, making proposals. Next on my list is Canada, to be followed by Italy, Sweden, and Spain. But first, the distinguished representative of Canada, so you have the floor.

Thank you, Chair. My delegation offers the following comments on box 4. We generally feel that this box is in good shape and that we should concentrate on actions that could assist States in meeting their international legal obligations. On paragraph 1, the first sentence is concerned with the legal review of new weapons means and methods of warfare under Article 36 and Additional Protocol I. Given that this is a legal obligation, the first sentence can be moved to box 2, while the second sentence, which is not a legal obligation, can remain in box 4 as a measure. On paragraph 2, while we agree that context-appropriate human judgment and control should be present throughout the weapons life cycle of laws, we find the current formulation somewhat ambiguous. We interpret the issue as the need to ensure that context-appropriate human judgment and control is considered at the earliest stages of design and development, not that human judgment and control is required during the design phase. An alternative formulation might be expressed in the following wording: The research, testing, and development of laws should consider context-appropriate human judgment in the activation and use of laws. On para 3, we have no concerns. On para 4, we agree with the delegations from Ukraine, Republic of Korea, France, Japan, and others for the addition of the word "significant" in front of "impact," as a significant change to the weapon would trigger an additional weapons review. However, minor changes would not necessarily require such a review. On paragraph 5, no concerns. Moving to paragraph 6, we support the notion of unintended bias, as bias is not always negative and may be necessary to discriminate targets. My delegation believes as well that an additional paragraph is necessary to address the considerations of IHL and ethics in the development of AI algorithms integrated in laws. This is necessary in order to avoid the presumption that the use of AI in laws will improve capacities, namely speed and accuracy, and therefore automatically improve civilian protection. We support the content of paragraph 7 and 8, but consider that they should be moved to box 5 as a measure which relates to the accountability and responsibilities of states. And finally, for paragraph 9, we agree with the concept of ensuring that controls exist for the transfer of non-compliant laws. However, we do recommend that the wording be reformulated to include an exception to accommodate the study of such weapons for the purposes of developing countermeasures and to facilitate the destruction of such weapons. I thank you.

Let me thank the distinguished representative of Canada for sharing the views of his delegation and making proposals. Next on my list is Italy, to be followed by Sweden. But first, the distinguished representative of Italy, you have the floor.

Thank you, Chair, for giving me the floor and allowing me to comment on this section. Our comments are going to be brief, as the language in this part of the rolling text is, in our view, rather solid. In the chapeau of this section, we would recommend replacing "should" with "must". This amendment is aimed at strengthening States' responsibility prior to the potential use of laws, excluding any margin of optionality that could otherwise be implied by "should". Additionally, in our view, should would appear to conflict with the rigorous test proposed later in the text. In para 1, we welcome the reference to the requirement to conduct legal reviews in accordance with states' obligations under international law. We would nevertheless recommend, as was stated previously by other delegations, replacing the term "deployment" with "employment," as employment more accurately reflects the focus on the actual use of a weapon, which is not just a mere fielding or positioning. Also, in para 8, we would prefer to change "human users" with "human operators." Finally, we would recommend to add a sentence to paragraph 8 that would reflect point F of the guiding principle, and that we consider a sort of security clause for the development and use of LOS, which is when developing or acquiring new weapons systems based on emerging technologies in the area of LOS, physical security, appropriate non-physical safeguards, including cybersecurity against Akin or data spoofing, the risk of acquisition by terrorist groups, and the risk of proliferation should be considered. And I thank you, Chair.

Let me thank the distinguished representative of Italy. And next on my list is Sweden, to be followed by Spain and Austria. But first, the distinguished representative of Sweden, you have the floor.

Thank you, Chair, and good afternoon, everyone. Sweden wishes to express support also for this box. We believe that it includes measures that are crucial to enable and ensure respect for and compliance with international law, in particular IHL. We have only smaller suggestions. First, we consider that this box contained both reflections of existing IHL and proposed practical measures that goes beyond existing law. We think that this should be highlighted, i.e. by placing para. 1 and para. 8, if para. 8 is refined to be more in line with existing IHL, before the current chapeau. Alternatively, we could support having a "shall" chapeau for para. 1 and 8, and a "should" chapeau for the rest. We support the changes made to paras. 4 to 8, including unintended bias, and would suggest a merge of paras. 6 and 7. We remain open to further refinement of how these paras. should be formulated. For example, we think that "sufficiently" could be added before "understood" in para 3, and that "riders" in para 5 could be deleted. We appreciate the spirit of para 9, but think that it requires more reflection. Sweden considers that laws that cannot be used in compliance with IHL should be prohibited. But if such systems appear and are used in violation of IHL, it might be necessary to deal with such systems, for example, to examine international crimes or develop capabilities to counter the threats such systems could pose. I thank you.

I thank the distinguished representative of Sweden for making proposals. Next on my list is Spain, to be followed by Austria and Norway. But first, the distinguished representative of Spain, you have the floor.

Thank you very much, Chair. We'll be very brief in our comments. We believe that this box, box four in the ruling text, includes paragraphs that are very well drafted, and we thank you and your team for this version of this box 4 in connection with the comments that were made by other delegations about the fact that box 4 refers to obligations that already exist in IHL. And we can join and support those delegations that request the replacement of the word should with the word must, both in the chapeau and also in paragraph 8, after the word states, states must ensure. In paragraph 2, For my delegation, it is important to stress the relevance that we already see in the chapeau, the fact that we are talking about the entire life cycle of these systems. So what we're trying to do is to ensure human judgment and control that is appropriate throughout the entire life cycle of the system. And perhaps we can add the word "entire" in paragraph 2. We already see it in the chapeau. And then paragraph 2 will read.

"Ensure there is context-appropriate human judgment and control across the entire laws lifecycle.".

And with that, we are ensuring that we're talking about all the phases and stages of the life cycle of the system. We also join other delegations that suggested for paragraph 4 the addition of the word "significant." And those will be our comments for the time being. Thank you.

I thank the distinguished representative of Spain. Next on my list is Austria. So you have the floor.

Thank you very much. I think we concur with others that this is the box where we are very close to actually closing one of the boxes. And finishing our work. We have very similar remarks as other delegations. I think the first thing that strikes us is also that some of the measures that are lined out here are actually obligations for some states. So there is still this problem that should in the chapeau does not accurately reflect this. We had some very interesting proposals. We could support to change the should to a must, but we also hear other proposals that could equally work. So I think there just needs to be a solution that would especially take the legal reviews out of this section, but we could also agree with the shell solution proposed by Canada. There are various ways on how to deal with this. Then there, we also see a similar problem like Canada when it comes to para 2, that it's not completely clear what is meant by this. And I think the idea originally was that the concept or the possibility to maintain context appropriate human judgment and control is considered or is taken into account for the whole life cycle, which means specifically or mostly the design development phase. We like the Canadian proposal. Our proposal would have been something along the lines to that ensure that the need to maintain context appropriate human judgment and control is considered across the, uh, laws life cycle, but it's, it's just one alternative to the proposals that we already here to address this. Um, on the other paras, we have just one comment for para 7, uh, which is about automation bias. Uh, it's extremely important for us that, uh, bias and, uh, automation bias are separated since it's, uh, two very, distinctive phenomenons and that in reality don't have much to do with each other. So what we also would suggest is that we use different verbs when it comes to the measures that states have to implement. So while detect, correct, or mitigate, or and mitigate, we are also looking favorable here at these proposals, might be important for bias, algorithmic bias, At seven, when it talks about automation bias, we would prefer that we talk about prevent, which in our view is the better word for using this. Just a quick remark, I mean, we still consider unintended bias to be a tautology, but we have kind of accepted that delegations like this tautology for safeguard reasons, but just to put it on the record again. Um, on eight, just a question, if this is intended that we are having a different, uh, beginning of this para, since, uh, all the others start, uh, with implement, ensure, and this is the only one which, uh, which also repeats the chapeau, maybe this is just a, a small mistake, and then finally, we would, uh, propose… a new para 8 bis, which would address the issue of data. We have already made this proposal during the intersessional period, and I will read it out. The new para would read, "Ensure the integrity, quality, and veracity of data used in the development and training of laws." And in addition, and this is now really the final point, we would also be interested in proposals by Ukraine and Italy to add new paras, specifically those who relate to security, cyber jamming, spoofing, and other issues which go into that direction. Thank you.

I thank the distinguished representative of Austria. Next on my list is Norway, to be followed by Cuba. But first, the distinguished representative of Norway, you have the floor.

Thank you, Chair, and thank you for the updated text, which includes many improvements based on the good discussions in March. We also feel that we are fairly close to some general agreement on this one, at least with its current content. Regarding paragraph 1, we very much welcome the inclusion of this subject, but one concern that we still have, and which we seem to share with a few States, Canada, Sweden, Spain and others, which also relates in part to paragraph 8, is about the relation between the chapeau and paragraph 1, which suggests that states should conduct legal reviews, whereas the first sentence of paragraph 1 reflects the language of what is, for many of us, the legal obligation contained in Article 36 of Protocol 1. So we understand and agree that the chapeau should remain phrased in a non-prescriptive way in light of the varied kinds of norms that are expressed in the nine paragraphs that follow. So we propose to fix this issue either by moving the first sentence to box two, like Canada proposed, or alternatively by relating the recommendation of the chapeau to the voluntary sharing of best practices, so the second sentence in that paragraph. So it would then read like this: "Engage, as appropriate, in the voluntary exchange of relevant best practices between States in order to facilitate legal reviews in accordance with their obligations under international law to determine in the study, development, acquisition, or adoption of a new weapon, means, or method of warfare whether its employment would, in some or all circumstances, be prohibited by international law." So basically changing the order. And that way, we break the link between the which should in the chapeau and the obligation to conduct legal reviews, whilst we still keep the emphasis on the importance of legal reviews, which is then further reinforced in paragraph four. So I will send this proposal to you by e-mail as well, of course. Also, we would like to express our appreciation for the new approach to bias, which clearly separates between data bias and automation bias. We think the proposals made by, I think it was Mexico and Australia, to add the word "and" would improve the text so that paragraph 6 and 7 would read detect, correct and mitigate. On the point that Australia raised that laws could rely on capabilities that are susceptible to data bias other than just artificial intelligence, we agree that we should aim as much as possible to make the measure as technology neutral as we can. Uh, if we do want to retain the emphasis on artificial intelligence, which currently perhaps is where we see this concern raised the most, we propose an alternative, which would read, uh, unintended bias in artificial intelligence or other capabilities relied upon by the laws. Um, yeah, and on the revised paragraph eight, uh, apart from the concern that I raised initially. We think this explains in a much more concrete way what it is we want to achieve with this training and to whom it should be directed to, so we support the direction you've taken this in. And finally, we appreciate the inclusion of paragraph nine, which we think is a very useful measure to avoid the weapons which it would be illegal to use are put into mass production and subsequently stockpiled, because if you have something in your stockpiles, there is always the risk of use. So this measure helps to avoid that, um, and by focusing on the manufacturing stage as opposed to the development and testing phase, we also resolve the problem that some mentioned in March, uh, and earlier that, um, at the development stage, you might not necessarily know what the weapon will end up looking like and that you might also have the need to develop and test prototypes for various reasons. So we think it should be considered to make this more than a recommendation and to co-locate this with the prohibitions and regulations in box 3 instead. Thank you.

Let me thank the distinguished representative of Norway and now pass the floor to the distinguished representative of Cuba, to be followed by Ireland.

You have the floor.

Thank you very much. Chair, I'd like to share with you some elements on paragraph 1. As some delegations mentioned already, in Article 36 of AP I to the Geneva Conventions, it clearly states that when a high contracting party studies, develops, acquires, or adopts a new weapon systems or new methods or means of war, it has the obligations to ensure that in some or all circumstances, it would not be a weapon that is prohibited by international law. This analysis has to be done also in cases of laws and also new, other new technologies for military purposes. We believe that national reviews of new weapons systems are not in itself sufficient measures because of the autonomous nature of such weapons, it's very difficult to attribute responsibilities to states or individuals when these weapon systems are used. in order to comply with norms of international law. Additional regulations are necessary for autonomous weapons. National reviews should be understood as a complement to a specific international norm that is necessary for this new category of weapons. The lack of uniformity and approaches related to national measures related to new weapon systems and elements reinforces the need for a specific international norm on autonomous weapons. We believe that that can be done through a legally binding instrument. On paragraph 6, we share the concern, the possible integration of AI in laws, taking into account the potential risks and biases during that use, where we need human control through the entire life cycle of these weapons, including critical functions. Machines cannot replace human beings in most important decisions related to war, and they cannot decide on the life or death of persons. On paragraph 7, my delegation believes that the focus should focus on lethal autonomous weapons systems, rather than simply the mitigation of risks. We must avoid them, and for that, we need to ensure human control. Now, on the last Paragraph we praise the prohibition on the production, acquisition, stockpiling of laws and also regulations are necessary for semi-autonomous weapon systems. Thank you.

I thank the distinguished representative of Cuba. Next on my list is Ireland, to be followed by the ICRC and Panama. But first, the distinguished representative of Ireland, you have the floor, sir.

Thank you very much, Chair. We see this box as reflecting the measures that should be taken to ensure compliance with IHL alongside context-appropriate human control. We reiterate, as others have done, that context-appropriate human judgment and control is not sufficient, but it is necessary to ensure compliance with IHL. We welcomed in particular the excellent contribution by Pakistan this morning. which eloquently analysed the task given to us by our mandate in ensuring international law is up to date and practically applicable to the demands of the realities posed by these emerging technologies, both articulating, adapting and applying existing rules of IHL and practices, and developing new principles where it is required. We see the measures in this box to sit in the former category, adapting and honing existing rules and practices to the particular challenges posed by laws. We also see these measures as being framed by the overall concept of being context appropriate and, as the chapeau text in this box says, to be applied as applicable. On that basis, overall, we see the measures in this box as being an excellent basis for further discussion towards an instrument. Two points on the chapeau. First, we welcome the fact that the measures apply across the life cycle. Second, we note the drafting tension, as others have flagged, in the use of the word "should" in connection with some of the paragraphs that sit beneath, in particular where they relate to the existing obligations of states. For example, the first sentence of paragraph 1 covers legal reviews, which form part of existing obligations under article 36. We noted both Canada's and Norway's contributions on this subject about moving the first sentence. They are constructive suggestions. As further examples, we think that paragraph 2 on context-appropriate human judgement and control, paragraphs 6 and 7 on bias, paragraph 8 on guidance and training and paragraph 9 on manufacture, acquiring, stockpiling or transferring should all begin with the word "must". We also note with interest the Mexican suggestion of requiring periodic reviews as a way of ensuring that developing technologies are dealt with by continued reviews, including by the virtue of machine learning. Chair, I'd now like to turn to paragraph six and seven on bias, and we welcome the continued focus on this very important topic. In 2024, Ireland, along with a group of states, submitted a working paper on addressing bias in laws. As that paper stated, algorithms and related machine learning risk repeating, amplifying, or contributing to unjust biases that programmers may not be aware of, or that are the result of narrow data selection. Database systems reproduce existing inequalities, and poorly programmed algorithms could mean that individuals are incorrectly recognized, or fail to identify properly civilian populations. This could fundamentally skew assessments of distinction, proportionality, and precaution in attack in line with IHL. We also note with interest the very recent work done by SIPRI on this issue, which recommends further work to elaborate on what IHL requires in terms of addressing bias. We think that this would be a valuable strand of work as we work towards a future instrument. As the Ukrainian delegate said, appropriate and thorough measures are basic precaution in the use of laws on this issue, both in ensuring their utility, but in particular, ensuring their compliance with IHL and the development and use of laws throughout their life cycle. We've heard two interesting, a few, but two in particular interesting and useful suggestions on this paragraph from Mexico in adding and or to ensure that it's not an exhaustive list, and from Australia in paragraph six to compel action where bias has been detected by adding the word and before correct or mitigate. As a further suggestion, we think that the phrase "as much as possible" is not necessary in either paragraph, as we believe that these would form part of the measures that states must take and so could be removed, as well as being subject to the overall caveat of being context appropriate. As a last point on bias, we again note the contribution by Ukraine about the importance of guidance and training. And so we read the words in paragraph nine apologies, paragraph 8, a proper understanding of the capabilities and limitations of laws as covering appropriate guidance, training and instruction on the possible effects of bias. Finally, we welcome paragraph 9, and we acknowledge your repeated exhortations that we should not be too caught up in the placement of items and which box they go in. We think that that element should sit in box 3, alongside the other prohibitions, as that would be logically consistent. Thank you.

I thank the distinguished representative of Ireland for making proposals and suggestions. Next on my list is the ICSC, to be followed by Panama, the Philippines, the UK and APILS. But first, the distinguished representative of the ICSC. You have the floor.

Thank you, Chair. We welcome the reference in paragraph 1 to the requirement to conduct legal reviews in accordance with states' obligations under international law. The ICRC views legal reviews as a critical measure to ensure that the use of autonomous weapons will not be in violation of IHL and other applicable international law. Given the mandatory nature of this obligation under Article 36 of Additional Protocol I to the Geneva Conventions, We would echo the comments already well made by some delegations, including Ireland just now, the change should, as currently seen in the chapeau, to must, at the very least for paragraphs one and four, although ideally also for the remaining paragraphs. We would also recommend replacing the term deployment with employment, as employment more accurately reflects the focus on the actual use of a weapon, means, or method of warfare, rather than its mere fielding or positioning. We would also suggest broadening the scope of the paragraph to cover modifications that render an existing system as a lethal autonomous weapons system as well. For example, it could read as, "States must conduct legal reviews to determine, in accordance with their obligations under international law, in the study, development, acquisition, adoption, or modification of an existing weapon system in a manner that renders it as a lethal autonomous weapon system, whether its employment would, in some or all circumstances, be prohibited by international law. We are open to other proposals that capture this point, including whether in another paragraph or as part of paragraph four. Namely, that a legal review, or potentially a re-review, must be carried out when an existing weapon system is modified in a way that gives autonomy in selection and engagement functions. We also welcome the reference in paragraph seven to automation bias and the separation of paragraph six and seven as requested by several delegations during the last GGE session. However, we suggest broadening the scope to cover harmful cognitive biases in human interaction with laws, including but not limited to automation bias. For example, this includes action bias, that is the tendency to prefer taking action over inaction even when unwarranted, which can equally distort human judgment and decision making. A formulation such as "mitigate harmful cognitive biases in human interaction with laws, such as automation bias," would more accurately capture the range of risks and better support compliance with IHL. We also echoed the suggestion made by Australia, Norway, and other delegations regarding changing paragraph 7 to read "detect and correct or mitigate." Thank you.

I thank the distinguished representative of the ICSC for presenting the views of his delegate for her delegation. Next on my list is Panama, to be followed by the Philippines. Madam, you have the floor.

Thank you very much, Chair. On the chapeau, my delegation agrees with speakers who took the floor before us to prefer the replacement of the word "deployment" with the word "employment" and to replace "should" with the word "must". We believe that this will be more appropriate language also in other paragraphs. Now, paragraph 1. This reflects a legal obligation under Article 36 of AP1. So we would prefer in the chapeau to use the word "must", and we share what was said by other delegations that in this case it would be more appropriate to replace "deployment" with the word "employment". We also can't support the proposal that was recently made by the ICRC in this paragraph to add the references to modification. On paragraph 2, if I'm not mistaken, Spain made a suggestion to add the word "entire", "entire life cycle". On paragraphs 6 and 7, we would be pleased to have a reference here to bias as Ireland mentions past year a group of countries including mine presented a document on algorithmic bias and also this was done as part of discussions on autonomous weapons and we support the suggestion that was made by Mexico supported by other countries to replace and or to say to correct and mitigate and do that in both paragraphs. And finally, we're ready to support the proposal that was made by Italy and Ukraine additional paragraphs that all refer to the risks represented by the systems and their vulnerability to cyber attacks, and that these weapons systems may end up in the hands of non-state actors. Thank you.

I thank the distinguished representative of Panama. And next on my list is the Philippines, to be followed by the UK. But first, the distinguished representative of Philippines, you have the floor.

Thank you, Mr. Chair. We share other delegations' view that box 4 is in good shape and is most likely to gather consensus sooner. On box 4, paragraph 1, our delegation supports the proposal to include the conduct of periodic legal reviews on the system of modification of laws. With increasing speed of technology, it is important to assess and evaluate from time to time if the employment of these weapons comply with IHL obligations. This proposed addition would also be consistent with paragraph four of the same box. Also, we propose that paragraphs four and five be reordered to follow paragraph one in box four, as we view that the said two paragraphs are closely related and provide concrete elaboration for paragraph one. This will allow a more concise sequencing of the measures. We also propose that the current paragraph 8 be reordered to follow current paragraph 3, for the same reason that we view that paragraph 3 and 8 are closely related and refer to each other. This is all from my delegation for now. Thank you very much, Mr. Chair.

I thank the distinguished representative of the Philippines. Next on my list is the UK. You have the floor, sir.

Thank you, Mr. Chair, and may I just take this opportunity to welcome Czechia to the group of states that have joined the draft articles on prohibitions and measures relating to autonomous weapon systems. Mr. Chair, we share the view of many in this room that this box is looking good and it's really getting to the material measures that may be used to address the questions that this group has got to address. We have a couple of points that we'd like to make and also to amplify the points that others in this room have made. Like others, we think that in paragraph 1, the use of the term "employment" is both more fitting and also more reflective of the legal test that we have in Article 36, which uses the term "employment." In relation to paragraph 2, we would support the points that were made by Canada and Austria and the drafting suggestion that was made by Canada in relation to paragraph 2. Paragraph 3, Mr. Chair, speaks about the capabilities and limitations of laws in their expected use and anticipated effects in different circumstances. We would suggest here that rather than the rather broad and undefinable different circumstances, formulation along the lines of all potential circumstances of use might better ground that paragraph. In relation to paragraph 6, Mr. Chair, which refers to algorithmic bias, we agree with the Australian suggestion, which both on drafting to include an and between detect and correct and mitigate, i.e. find and then fix. We would also agree with the Australian point in relation to the breadth of the article being narrowed by using artificial intelligence capabilities there. We would add that you could add after unintended or unanticipated bias. So altogether, Mr. Chair, that paragraph would read, implement measures to detect and correct or mitigate as much as possible unintended or unanticipated bias, including bias in data sets and AI models. On number 7, which refers to automation bias, which concerns the behaviour of the users and the humans, I note the relationship between the concept of automation bias in paragraph 7 and some of the measures in paragraph 8, which may well be measures that you would deploy in order to address the issue. We agree with the point made by Austria as to how we might wish to, in fact, be looking to prevent automation bias, and we might be deploying some of those measures in paragraph 8 to do so. However, we would suggest that rather than just use the term "prevent", you could use the terms "all reasonable steps to prevent". So therefore, Mr. Chair, the paragraph would read, "Implement measures to take all reasonable steps to prevent, as much as possible, automation bias." Thank you, Mr. Chair.

I thank the distinguished representative of the UK. Next on my list is AP ILS, to be followed by France and Uruguay. Sir, you have the floor.

Thank you, Chair. This box contains an important set of practical measures to be taken across the life cycle of laws to improve compliance with international law. It is good to see general support for these kinds of measures and the numerous suggestions for improvement. The chapeau of this box concludes with the word "should", indicating that the paragraphs that follow reflect desirable measures or good practices. But this creates issues in relation to paragraphs 1 and 9. As several delegations have underlined, the obligation to undertake legal reviews of new weapons, reflected in the first sentence of paragraph 1, constitutes a legal obligation at least for those States that are party to Additional Protocol I. Stronger language from the Group's earlier consensus documents would be preferable to reflect this. For example, the language using the word must from paragraph 23 of the 2023 report seems more appropriate. The second sentence of this paragraph can remain the same. And to accommodate the stronger language, paragraph one could be brought to the very beginning of the box before the current chapeau. But we also find the proposal to move the first sentence of paragraph one to box three interesting. But if that is done, the second sentence would need to be rephrased here to bring it in line with the Chapeau. Paragraph 9 creates a different issue in light of the Chapeau. We do not think that general IHL can be interpreted as prohibiting the manufacture, acquisition, stockpiling, or transfer. Hence, this paragraph appears to propose a new requirement by using the words states undertake. We have no objection to this proposal as such, but the words "states undertake" in this paragraph clash with the words "states should" in the. Conversely, in paragraph 8, the words "state should" seems to be redundant as they duplicate the same words in the. Chair, turning now very briefly to the substance of the list of measures. We support the intent of paragraph 3 to ensure that the capabilities, effects and limitations of laws are understood. But the current language seems to create impossible requirements for states, especially during legal review. States cannot be expected to anticipate all of the effects of laws in all conceivable circumstances of use. To address this issue, the words "in different circumstances" in paragraph 3 could be replaced with the words "in other reasonably foreseeable circumstances of use." In paragraph 4, changing "modification" to "significant modification" appears to be a sensible change. The idea of conducting periodic reviews is interesting. but raises a host of questions about the appropriate timing of such reviews. Conducting reviews after significant modifications appears to be clearer and more in line with the existing law. Finally, we would also support the proposed changes to create a stronger connection between the legal review obligation in paragraph 1 and the flow on requirements in paragraph four by locating these paragraphs together. I thank you, Chair.

I thank the distinguished representative of IPILS, and next on my list is France, to be followed by Uruguay and Brazil, but first the distinguished representative of France, who's available?

Thank you, Chair. In our joint non-paper, we explained how the notion of context-appropriate human judgment or control is applicable to the development phase. During that development phase, human control can be exercised, for example, by guaranteeing that the target identification criteria are compatible with IHL. Humans must also in advance evaluate and assess the feasibility, the comprehensibility and the predictability of these weapons systems. For example, through appropriate evaluation test processes, allowing us to assess or evaluate how these systems would function in use case scenarios, on predicted use case scenarios. With regard to our comments on the draft rolling text and box 4 thereof, on the chapeau, to echo a point made by Sweden, we need in this section to make a distinction between those paragraphs that stem directly from obligations under IHL and therefore need to be preceded by the use of the word must in order not to reduce the scope of those obligations which are pre-existing under IHL, but we need to distinguish those from other paragraphs that don't stem from IHL and which should be prefaced by the use of the word should. In paragraph 3, we would like to add the adjective "adequately" before the word "understood" at the very end, or "sufficiently" as an alternative, as proposed by Sweden, so "are adequately understood" or "sufficiently understood". This would mirror the language already used in 6. In paragraph 4, here we echo other delegations that took the floor before us, such as Japan, we'd like to add once again the word "significant" before the word "impact". Any change does require... Not any change requires a legal review, only those changes that significantly modify the characteristics of the system. Moreover, this paragraph could be moved up close. Could be rendered closer to paragraph 1 along the line suggested by the Philippines. In paragraph 7, We don't want to go back to the use of the wording harmful bias because not all forms of bias are necessarily harmful. So we would prefer to avoid using harmful bias, but perhaps we could use a more neutral adjective such as adverse, adverse bias or disruptive, disruptive bias, but not harmful. In paragraph 8, we would like to replace the word "proper" with the word "sufficient" before the word "understanding". The idea being that the knowledge of the system and their impact by those responsible for those systems be sufficient to fulfil the legal requirements linked to their responsibility. Finally, paragraph 9, we think doesn't fit properly in box 4, so it could be deleted, all the more so because we believe it is redundant because some of the elements contained in paragraph 9 have already been referred to in box 3. So we suggest deleting the current paragraph 9, but we could support the inclusion of a new paragraph on avoiding proliferation of non-agreed actors using already agreed language to capture that concept. Thank you very much.

I thank the distinguished representative of France. Next on my list is Uruguay, to be followed by Brazil and Singapore. But let me first pass the floor to the distinguished representative of Uruguay. You have the floor.

Thank you, Chair. We would like to support the replacement in the chapeau. of the word should by the word must or at least use should for those paragraphs that refer to obligations under IHL or not those paragraphs rather. We agree with Spain's proposal with regard to paragraph 2 because we believe this analysis needs to cover all the elements referred to the deployment, use, modification, acquisition, finalization of the life cycle of the weapon. We support the inclusion of rigorous legal review measures and a rigorous approach to bias. We think that's duly captured in the text, but we believe that we need to bolster this concept that responsibility and attribution or accountability for the use of these systems falls not on the machines themselves, but on the state parties involved in a conflict and on the individuals that are operating those systems at each step of their life cycle. Thank you very much.

Let me thank the distinguished representative of Uruguay. Next on my list is Brazil, to be Singapore, and Singapore is for the time being last on my list. But first, the distinguished representative of Brazil, you have the floor.

Thank you, Mr. Chairman. On this box, my delegation would like to offer a few comments on proposals that we have heard and some proposals of our own, beginning with the chapeau. We would like to express our support to the proposals made by Japan and Korea regarding the need to reflect the importance of protecting the right to peaceful uses, although my delegation has an inclination to suggest that this reference is included in box 2 as it is or should be an overarching principle. We also support the expressions of the importance of reviewing the use of the word "should" here. I think that many interesting ideas have been put forward in this regard to readjust this language throughout the subparagraphs in accordance to where the best word would be applicable with regards to paragraphs that refer to legal obligations. Turning now to paragraph, sub-paragraph 1, my delegation would like to concur with the proposal made by Canada with regards to the possibility of splitting this paragraph in two, but we would prefer both of them to be kept in this box. With regard to the last sentence, the one that begins with, "In this context," we would like to propose the inclusion of reference to the possibility of developing technical standards for this process, while recalling the proposal made by Brazil last year and updated this year on the possible contribution of IEEE technical standards for the work of this GGE. With regard to paragraph two, my delegation also heard with interest the proposal made by Canada with a view to making this formulation more clear and direct. We would certainly be interested in that fix. On paragraph three, my delegation supports the proposal made by Ukraine to include at the end after understood, a reference to the chain of command in order to clarify to whom this responsibility refers. On paragraph 6, my delegation supports the amendments proposed both by Australia and Mexico with regards to the terms "detect, correct, and mitigate." I think that These amendments make the text clearer as regards the fact that these are not alternative references, but complementary. We also propose on paragraph 7 that the same amendments are made, also taking into account Austria's suggestion to replace "mitigate" with "prevent". We also support the proposal made by the CRC to make more specific reference to what we mean by automation bias, which would be cognitive bias in human interaction with laws. This makes it even more user-friendly to the readers of this text that this is a different type of bias than the one that we're talking about in paragraph 6. On paragraph 8, actually not on paragraph 8, the Ukraine's proposal on a new paragraph on the security aspects, including jamming and spoofing, is also an interesting one. And finally, on paragraph 9, we also concur with the suggestion by Ireland to move it to box 3, as it probably is better placed there, right after the paragraph on development. I thank you, Mr. Chairman.

I thank the distinguished representative of Brazil. Next on my list is Singapore, to be followed by the United States. But first, the floor for the distinguished representative of Singapore.

Thank you, Chair, for giving me the floor. My delegation is of the view that box 4 is a crucial aspect of the rolling text. The national-level governance measures contained in box 4 help to translate broad international law obligations into practical steps and reduce the risk arising from misuse or accidents involving laws. We support the explicit reference to the conduct of legal reviews in box 4. Legal reviews serve as a useful tool for countries to determine whether the use of specific weapon systems would, in some or all circumstances, be prohibited by international law. My delegation notes that paragraph 4 seeks to clarify when a subsequent legal review would be necessary during the life cycle of laws. The paragraph currently states that any modification would trigger a further legal review without regard to the nature, extent or type of modification. To be clear, Article 36 of AP I to the 1977 Geneva Conventions is silent on the need for States to conduct multiple rounds of legal reviews in respect of every incremental development or modification to existing weapons. That's it. In the spirit of being constructive, we are open to considering the addition of language specifying that changes would have to be significant in order to trigger another legal review. This is in line with proposals made by colleagues from Canada, France, Japan, the ROK, Spain and Ukraine. As a broader observation, box 4, paragraph 4, as well as a few other paragraphs in the other boxes, appear to be scoped to focus on only the functions of select and engage. Mind allegation is therefore concerned that the omission of the critical identify function would have implications for how the measures in these paragraphs apply. Let's take paragraph 4 as an example. if significant modifications are made to the target recognition module of a hypothetical laws, it may not trigger another round of legal review because target recognition specifically pertains to the identify function. My delegation is therefore of the view that this meeting should further consider and discuss the risks that may arise from specifying only the functions of select and engage in these paragraphs. I thank you, Chair.

I thank the distinguished representative of Singapore for sharing the views of his delegation and making proposals. Next on my list is the United States. You have the floor.

Thank you, Mr. Chair. The United States delegation thanks Czechia for its statement and welcomes it to the group of states supporting the draft articles on autonomous weapons systems, which has been resubmitted as a working paper to reflect their additional co-sponsorship. We believe the proposal has great value in clarifying how existing cardinal principles of international humanitarian law apply to the development, deployment, and use of autonomous weapons systems. And we encourage other delegations to consider its substance and to endorse it. Next, we support the Republic of Korea's proposal to add language from Guiding Principle J to the chapeau of box four. We like how simple the chapeau is. as we think that simplicity will help us reach consensus on the language. But we also think it would be useful in the chapeau of Box 4 to link these paragraphs to the prior paragraphs. We also think it would be useful to use the singular here to clarify the responsibility of each state. And we think "throughout" should be replaced with "in" because some of the measures are taken in specific parts of the lifecycle rather than throughout the entire lifecycle. So the chapeau would read, in order to uphold compliance with the prohibitions and restrictions set forth in the preceding paragraphs, prior to deployment and as applicable in the entire lifecycle of its laws, a state should. Regarding paragraph one, first we propose moving the phrase in accordance with their obligations under international law to the beginning of this paragraph. This might make it more clear that this is an obligation for some states rather than a should. Second, we recommend omitting the phrase in the study, development, acquisition, or adoption of laws. The chapeau already has this language as applicable in the entire life cycle. and the phrase in the study development, et cetera, can also be understood to be incorporated into the phrase in accordance with their obligations under international law. We note the second sentence of this paragraph could be considered as another possible measure to recommend to CCW high contracting parties. Lastly, we propose to change its deployment to the state's use of a new laws. This would make it more consistent with Guiding Principle E and the 1977 Additional Protocol I, which relate to new weapons. Some laws are not new weapons, but are existing weapons, and there will of course be a time when a new law is no longer new, after a legal review has been conducted. So the paragraph would read, "In accordance with their obligations under international law, conduct legal reviews to determine whether the state's use of a new laws would in some or all circumstances be prohibited by the state's obligations under international law. So last point, we'd like to add the state's obligations because of course each state has different obligations depending on the instruments that it has ratified. Regarding paragraph two, my delegation has stated its general concerns with the terminology context appropriate human judgment and control, and many have asked what it means. We oppose creating unclear, overly subjective requirements that will lead to controversies because there is not a common understanding of what is required and whether those requirements are met. This paragraph also doesn't make sense to us because control occurs during the use of the weapon rather than across the life cycle. Instead, our proposal is as follows: take appropriate measures in the development of the laws to mitigate the risk of unintended engagements. This builds on one of the ideas in the working paper submitted by Bulgaria and other delegations, which note the importance of the intentions of commanders and operators. And Austria has reminded us this week about the importance of human agency in the use of force. In paragraph 23a of the 2019 report, the GGE found consensus on the need to take measures to mitigate the risk of unintended engagements. Regarding paragraph 3, we have some edits to it, but we recommend merging it with paragraph five below. Our main concern with this paragraph, paragraph three, is to relate it to specific IHL requirements rather than trying to develop a new abstract requirement based on understanding the weapon. Regarding paragraph four, we recommend simplifying this paragraph and drawing a logical connection with paragraph one. The review of the modification needs to occur when the modification makes it a new weapon. So we recommend the text read as follows. Ensure that any modification to the laws that would make it a new type of laws is reviewed. Next, we recommend including a chapeau that would cover paragraph 5, paragraph 6, and elements from box 3, paragraph 6. In box 3, we think it is useful to articulate the obligation to take feasible precautions in the use of the system. Here in box 4, we believe it would be useful to articulate measures in the development that affect the nature of the laws. Measures in the development of the system are often much different than the measures that can be taken in the use of the system. So we see a practical implementation benefit for having different lists to make our future instrument more effective. Regarding paragraph six, first we would omit detect, correct, or, because we see these as part of mitigating. Second, we would add that enable the laws to detect and pardon me, that enable the laws to select and engage targets after artificial intelligence capabilities. This paragraph needs to be related to laws and we should be focused on those functions that are most pertinent to IHL implementation. We also recommend adding such before bias and data sets and AI models. The paragraph would read implementing measures to mitigate as much as possible unintended bias in artificial intelligence capabilities that enable the laws to select and engage targets, including such bias in data sets and AI models. Next, we would place elements of box three, paragraph six here. IHL, in our view, doesn't impose obligations in development, but it is necessary for measures during development to be taken so that laws can be used in compliance with IHL. So we believe states should provide measures that enable the operator to anticipate and control the effects of using the laws or that otherwise control, limit, or otherwise affects the types of targets, duration, geographic scope, and scale of the operation of the laws, including through, as appropriate. So I won't repeat all of our specific measures and proposals here, but in response, we did want to respond to the debate about these kinds of measures. And in particular, we support incorporating limits or enabling an operator to control the number of engagements the laws can undertake autonomously. Here, I also wanted to respond to some of the questions that have been raised about what word we should use. Should we be using control or limit or other words? For us, we see control as occurring during the operation of the system and to involve deliberate decision making by the person. Control is active. For us, limits seems to be an uncontroversial idea. And we respectfully disagree with delegations that have raised an operational problem with the idea of limiting the number of engagements that LOS can undertake autonomously. For us, every LOS has a limited number of engagements that's defined by how much ammunition it is given or how much fuel it carries. But it is a good practice during development to consider the number of engagements the system can undertake autonomously to support compliance with IHL in the future use of the weapon. We also would use the phrase otherwise effect in some of our proposals. For us, this terminology reflects the complexity of human-machine interaction. Sometimes people engineer or encode strict limits in the system. Sometimes people control the system directly. Other times, people take actions that affect the system's operation, but not as a limit nor a control. So, for example, the choices made in selecting the data sets for a machine learning algorithm affect its operation, but are not a limit nor a control. Regarding paragraph seven and eight, we recommend moving them to box five because they relate to personnel and accountability. Regarding paragraph nine, we echo the concerns that have been raised about this paragraph. We would also note that for us, this obligation, kind of obligation or possible obligation, needs to be considered in light of the discussion on characterization. Autonomy is a function that can be present in a wide range of systems. The difference between an inherently indiscriminate laws and one that can be used lawfully could be in software code rather than readily apparent from the outside of the system. I thank you, Chair and other delegations, for your patience in listening to this long statement.

I thank the distinguished representative of the United States for stating his position, making comments and proposals, and also referring to a working paper that was previously issued. Next on my list is Belgium. Let me pass the floor to the distinguished representative of Belgium. You have the floor.

Thank you, Mr. Chairperson. On box four, we can be relatively brief, as it is in a very good state. We are open to positively considering the suggestions made by different delegations to break the link between the verb "should" and the legal reviews under paragraph 1. Furthermore, on paragraph 1, we support the proposal made by various delegations to change "deployment" to "employment" for the reasons given.

In relation to paragraph 4, we share the view that such a review indeed requires significant modifications.

On paragraph 9, we believe that it is important to encompass the development phase in this provision. We would therefore suggest the following wording. States undertake not to develop, manufacture, otherwise acquire, stockpile or transfer laws that cannot be used in compliance with IHL. As referred to by the French delegation, we see merit in addressing the issue of diversion in this section. Preventing terrorist groups, criminals and other unauthorized unauthorised actors from getting their hands on laws is a shared objective, we believe. The language we have previously proposed in this regard reads: "States should ensure the adoption of measures to prevent, halt and address the diversion of laws to unauthorised.

End users." I thank you, Chair.

Let me thank the distinguished representative of Belgium for also making concrete suggestions. Next on my list is the distinguished representative of Israel. You have the floor, sir.

Thank you, Chair. Regarding the first paragraph in box 4 of the rolling text, The current text does not adequately reflect the varying legal obligations of states. As currently phrased, it could be interpreted as implying that all states are equally obligated to conduct legal reviews. However, there are states that are not parties to the First Additional Protocol to the Geneva Conventions, and as such, are not bound by Article 36 of this protocol, even though some, including Israel, conduct reviews as good practice. For this reason, and in order to align the text more closely with the agreed language in Principle E of the 2019 Guiding Principles, we suggest the following text: "in accordance with their obligations under international law, in the study, development, acquisition, or adoption of laws, determine whether its deployment would, in some or all circumstances, be prohibited by international law." Regarding the second paragraph, with respect to the term "context-appropriate human judgment and control," we refer to our previous comments and ask to delete this paragraph. Regarding the third paragraph, in order to clarify the underlying point of the paragraph, we suggest adding the words "to its users" at the end of the paragraph so that it would read, "ensure that the capabilities and limitations of laws its expected use and its anticipated effects in different circumstances are understood to its users. Regarding the fourth paragraph in box four, we wish to note that not every development or modification even if it concerns the target selection and engagement functions or the anticipated effects, necessarily requires further review. For example, a reduction in the number of the system's permitted engagements or the replacement of the armament to one that does less damage does not necessarily require further review. For this reason, As proposed earlier by several other delegations, like the Republic of Korea, Japan, Canada, we consider it important that the text refers to developments or modifications that have a significant impact. Additionally, as this paragraph is also dealing with legal reviews, the caveat in paragraph 1, which is in accordance with their obligations, is equally relevant here, and we suggest adding it at the end of the paragraph. Therefore, it would read ensure that any further development or modification to existing laws that have a significant impact on their target selection and engagement functions or on its anticipated effects are reviewed to ensure compliance with IHL in accordance with their obligations under international law. Regarding the eighth paragraph, there seems to be an overlap between this paragraph and paragraph three, as they both focus on ensuring understanding of laws. In order to streamline the text, we would suggest making modifications to it in the eighth paragraph, that I will read, "States should ensure appropriate guidance, training, and instructions for human users of laws." Regarding the ninth paragraph, in our view, our focus should be on the use of laws and not on other actions regarding laws, such as manufacturing or acquiring, as they stray from the CCW's objective and expertise in the IHL-focused treaty. We therefore request that this paragraph be deleted. I thank you, and we will share this in writing as well. Let.

Me thank the distinguished representative of Israel for comments and proposals, and indeed, we appreciate to receive that in writing. Next on my list is the University of Nottingham. Madam, you have the floor.

Yes, thank you. I have a couple of points on box 4. Yes, article 36 of AP1 is an obligation in treaty but also customary international law that legal reviews of weapons or weapons systems are undertaken by states. One of the problems of this obligation is that it is not transparent in terms of sharing the outcome of reviews, or indeed whatever type of review has been undertaken by that particular state. There may be states which have particular arrangements, friendly arrangements, where they do share that information. but there is no obligation to release that to other States. That leads me on to the next issue. If that is the case, there is nothing currently in box 4 that prohibits a State from selling and exporting to another State that does not possess the technology, and that can go completely under the radar. and the state that then purchases the technology would have to undertake a review because it would constitute a new weapon, but that relationship would fall under the radar of the provisions that we have here at the moment. And it seems quite contradictory if in previous boxes we have referred to prohibitions and restrictions, if that's the case, that weapons or weapon systems are being sold or exported which cannot comply with IHL because they fall into the prohibitions and restrictions, then logically there should be some expression and reflection of that here in box 4. In paragraph 9 of box 4, this is a weak formulation at the end where it says that cannot be used in compliance with IHL. Arguably, the state that is selling and exporting these weapons or weapon systems may be able to argue their way that the state they're selling to is a matter for their jurisdiction, and as far as they're concerned, through their due diligence, they are aware that there is compliance with IHL. And those are some of the problems that I think are inherent here in the formulation for box four. Thank you.

Let me thank the distinguished representative of the University of Nottingham for referring to especially the issue of possible transfers or export. I recognize the distinguished representative of the Russian Federation. So you have the floor.

Thank you very much, distinguished Chair. We would like to share with you some of our ideas related to box 4 on paragraph 1. Overall, this paragraph does not create any big questions for us. The only thing that we'd like to do is to bring it in line with language that has already been adopted related to the need to conduct legal reviews, namely, instead of having a simple reference to lethal autonomous weapons systems, to add a reference overall to new weapons means or methods of war, including lethal autonomous weapons systems, in order to not create the impression that this obligation only covers lethal autonomous weapons systems. On paragraph two, just as In the case of box three, we see that some of the paragraphs of box four repeat some of the separate understandings that we find in previous boxes or are in box five or something that we already have in box four. On paragraph two, we believe that some of the previous provisions factually already cover the idea contained in this paragraph. And to be precise, of course, we are talking about paragraph 5 of box 2. And also I'm referring to some of the provisions contained in box 3 including paragraph 6 of box 3, and also paragraph 1 of this very box, something that we consider to be as one of the elements of human control. Altogether, all of those paragraphs represent the understanding that we currently have in paragraph 2. In other words, human judgment and control should be ensured throughout the entire life cycle of such weapon systems. This is why we propose the deletion of paragraph 2. On paragraph 3, we do not have any particular objections. We believe that this paragraph is very relevant, and should be included in this text as an important understanding in connection with such weapon systems. Moreover, we would like to stress the fact that this paragraph is directly linked to paragraph 4 of box 3, that is there, but we believe it should be presented more broadly on the basis of paragraph 23A from the 2019 report of the GGE. On paragraph 4, we believe this paragraph also represents a repetition of previous paragraphs. In particular, paragraph one of this very box, because paragraph one relates to legal reviews in general. And also this relates to possible modifications of weapons systems if such modifications are related to introducing such characteristics that will make such a weapons systems a new type of weapons system, a new means or method of war. Otherwise, We believe it would be very simply to have a primary legal review or an initial legal review to ensure that such a weapon system would fully comply with the norms and principles of IHL. And in that case, if such modifications are not really significant ones, the conduction of a new legal review will not be necessary. This is why we suggest the deletion of paragraph four. Paragraph five does not create any problems or questions for us, and we're ready to accept the language that we see here, the language that you have presented, Mr. Chair. We believe that it is very important to carry out testing and relevant evaluation related to the functioning of the system, especially as part of the different operational environments. This is something that is extremely important. On paragraph 6 and 7, as we said earlier, we do not see the need to make a separation between these two paragraphs and presenting them separately, we believe that it will be enough to have one single paragraph that will be related to different faults or errors of the system and different bias. And we believe that in this case, we can add the reference to automation bias. at the end of paragraph 6 as one of the particular situations related to such bias that will require relevant action or measures by member states, states parties. On paragraph 8, Here, we believe that we need to have some serious work on this paragraph. First of all, we would prefer the deletion of the word "appropriate," a word that may create possible different interpretations in the future as a characteristic that is subjective in its nature. since we believe that this will create problems with determining where is the level of relevance related to this paragraph. What will actually be appropriate and what will not be appropriate. Also, in this paragraph, We believe that it would be very important to be consistent in the terminology that we use. In particular, if we are talking about the chain of command, then we have to talk about chain of not simply command, but we have to talk about the chain of command and control. This is the term that we used earlier in the text, and we believe that it is very important to use it throughout the text in order to make sure that we're talking about the same thing. Also, paragraph 8, we would prefer to see the last part of the paragraph deleted that begins with the words "in order to" until the end of the paragraph. We believe that this language is superfluous. It doesn't add anything important. It doesn't have any practical purpose. It only relates to the motives of taking relevant measures and the inclusion of the relevant understanding of the text to be simply unnecessary. We should not be talking about what are the motives for including this. The most important thing is that we all agree that we must include such an understanding. So, we believe that we should sacrifice the last part of the sentence and fully delete the language that begins with the words "in order to provide". On paragraph 9, we find that this paragraph is also superfluous. from a substantive point of view, from a fundamental and conceptual point of view, we believe that such a prohibition is simply not necessary because the situation and activities that are covered by this prohibition, all of that in light of the other understandings that are included in this document, This is something that is simply impossible. Such activity cannot be carried out in practice, taking into account the fact that we have legal reviews that should be carried out or conducted. And it is at that stage is where we should determine and decide whether this or that weapon system is in compliance with humanitarian law or not. And if it is determined that the system is not in compliance with international humanitarian law, then that system simply will not be manufactured. And also, this means that such a system would not be acquired or transferred to anyone. One could not be stockpiling such a system due to the fact that during the stage of legal reviews, such a weapon system simply be with removed from the process of production or any other subsequent inclusion in the armed forces of a country. So this is why we suggest we simply delete paragraph 9. Thank you.

I thank the distinguished representative of the Russian Federation for stating the views of his delegation and making proposals with regard to the text. Next and now last on my list is the distinguished representative of Luxembourg. You have the floor.

Thank you, Chair, for giving us the floor. In general, we are very supportive of how Box 4 is currently drafted. We only have a few small additional comments we would like to make. First, regarding paragraph 4, as it again refers to reviews that should be conducted, and thus implicitly to Article 36 of Protocol I of the Geneva Conventions, as in paragraph 1, we would suggest either merging this paragraph with paragraph 1, or alternatively, moving it up in the text to constitute a new paragraph 2. This would make the text read more logically and clearly set out that legal reviews are required not only at the initial stage of developing and or acquiring a system, but also whenever the system is further developed or modified. We also support the comments already made by many other delegations to insert significant before impact in the current paragraph four for the same reasons they have eloquently outlined. Finally, regarding paragraph nine, we are not convinced that this is the right place to set out such an essential obligation in the rolling text. This is, in our view, a general obligation that does not necessarily fit under the chapeau of box 4.

We would therefore suggest.

Moving it to box 5 so as to avoid linking this obligation to the chapeau of box 4. Thank you, Chair.

I thank the distinguished representative of Luxembourg, and I now recognize the distinguished representative of Colombia. Sir, you have the floor.

Thank you, Chair, for giving us the floor. We will make very specific comments on box 4 of the rolling text on the chapeau. We join delegations that proposing the change of the word "deployment" and use the word "employment." And we also find it appropriate to change the word "should" and replace it with the word "must." On paragraph 1, we appreciate the inclusion of legal reviews. We join the proposal that was made by Mexico related to periodic reviews. And 2, we support the proposal made by Spain to refer to the entire life cycle of laws. that are legal. On 6 and 7, we join delegations that prefer to replace the word, to implement measures to detect, correct, and mitigate, replacing or with and. Also on 6, we support the proposal made by the ICRC and other delegations to replace the word unintended with the word harmful. On paragraph 9, we support the proposal made by Belgium to include the restriction of the development of laws. Thank you.

No more requests for the floor. Yes, I recognize Samoa. Madam, you have the floor.

Thank you, Chair. I'll be brief. In paragraph one, we kindly recommend adding testing after development. This is to complement what is reflected in paragraph five and for reasons that were conveyed at this morning's session. I thank you.

I thank the distinguished representative of Samoa. I think we have exhausted the list of speakers. Looking back at our discussion, I think we have consensus on doing some work with regard to power one, bring that in line with the different obligations of high contracting parties. Also the idea of a possible split, but indeed to also reflect, think about power four in relation to power one. In power two, we have an issue that we It's an old friend, I would say, by now, we had it in two, we had it in three, and we have it here, so, uh, on the other hand, if we solve it once, we might solve it three times, so, uh, I leave that for what it is, um, and we have further, uh, proposals with re- regards to seven and eight, um, we have a different appreciation of nine, that's clear, uh, and we have some proposals to possible mergers. and some detail. I think one of the things, for instance, with regard to para 4, the word "significant" has very broad support. So there is clearly some homework to be done by the Chair's team to clean up the text, and I thank you all for your suggestions. And as always, we would greatly benefit from receiving them in writing. We have all been making notes. This is how my notes look like. So you can imagine that it's quite a challenge to make something new out of that. But we do our best, of course. We have one more box to go, and we have only half an hour. But I think we should make best use of our time. And we will do so. We will keep you here for another half hour. I had a few speaking notes to open the discussion, so let's make use of them. And I think I can be brief. Can we put it up, box number five? Yes, there it is. In this box, the language has been reorganized and adjusted in response to numerous constructive inputs from delegations. Let me highlight that. For example, I added an overarching principle in para. 1 of box 5, in line with Guiding Principle Bravo of the 2019 GGE report about IHL obligations on human responsibility and accountability, so that should look familiar to you. I furthermore also inserted some new language in paragraph 5 to specify how to promote human responsibility and accountability. Overall, and that is perhaps a dangerous assumption, but let's try, I believe this box to be in a good shape and I'm hopeful that we can really move to clear it up, uh, if there are still some issues, uh, this week, so we don't have to continue reviewing that in the future, uh, but let me restate it again, nothing is agreed until everything is agreed, but, uh, let's see whether we can come to something that elsewhere is called generally accepted, um, uh, because we should really, uh, then, focus on issues where still a lot of work is to be done. Having said that, I would now like to open the floor for comments and suggestions. Who would like to kick off? I already see the plate from Ukraine. Thank you very much for that, for breaking the ice on it. And please raise your flags if you also have comments to pass, because then we can still do that this afternoon. But let me first pass the floor to the distinguished representative of Ukraine. Sir, you have the floor.

Thank you, Mr. Chair. The delegation of Ukraine does hope that our statement could serve as an icebreaker, although we don't expect this to happen because it's going to be pretty general. It will concern Box 5 in general, and we reserve the right to further address the particular elements of that box in more detail in later statements. The delegation of Ukraine generally welcomes the intention of determining the basic approaches towards responsibility for the use of force. However, we also call upon delegates to treat this issue with utmost caution. Here, the delegation of Ukraine would like to remind about the history of the development of the articles on responsibility of states for internationally wrongful acts. For decades, states keep negotiating about particular responsibility mechanisms. We warn against attempts to formalize the rules and responsibility of states at this forum prior to the completion of the relevant efforts by the International Law Commission. Many states may find it uneasy to adopt for themselves a duty to bear international responsibility, and this complexity may lead to the overall failure of our potential instrument. The delegation of Ukraine does not exclude the possibility of inclusion of certain provisions on responsibility, specifically pertaining to laws, but we are convinced that this GGE should carefully avoid any provisions that may potentially create more general rules establishing state responsibility. Thank you for your attention to this statement.

I thank the distinguished representative of Ukraine for making this initial statement. Let me also refer to, of course, what we have at the very beginning of the text, that this is all for the purpose of the work of the GGE, indeed, very specific to laws. We had earlier this discussion whether this is about general obligations or specific obligations. Oh, we have already two more. We have the United States and the Russian Federation. So let me now pass the floor to the US. You have the floor.

Thank you, Mr. Chair, for giving me the floor to share some of the US delegation's thoughts on box 5. At the start, I want to note that we agree with the suggestion made by Bangladesh earlier in the week to move some language on the accountability earlier in the document to box 2. The concept that humans must remain responsible and accountable was enshrined in paragraph B of the guiding principles, and we believe highlighting it in box two is appropriate. As for box five, we see the value of this section to be that it reaffirms the applicability of established legal principles of responsibility in relation to the use of laws in armed conflict. It also articulates measures specific to laws that strengthen accountability for their use. In this way, it is focused on measures that address people instead of the technical focus measures in box four. We have no edits to paragraph one. In paragraph two, we would delete the term "must" since those responsible for violations attributed to them are responsible as a matter of law. The use of the word "must" can give a misimpression that states are not already responsible for their decisions. We heard the Austrian concerns from this morning that there is a gap in international humanitarian law on accountability. We think that is not true, and we believe this edit helps reflect that fact. In addition, in paragraph 2, we recommend adding "their" before "decisions" to clarify that the States are not responsible for other States or parties' use of laws. We also recommend moving the notion in paragraph two of accountability, including that it cannot be transferred to machines, to the next paragraph, which already has a focus on accountability. The resulting paragraph two would read, "States, parties to armed conflict, and individuals at all times remain responsible in accordance with applicable international law for their decisions with regard to laws." We think paragraphs 3, 4, 6, and 7 are duplicative and should be combined into one paragraph addressing accountability. We also recommend removing the "means or methods of warfare" language from the first sentence in paragraph 3 and replacing it with "their use of laws" to simplify the text. The resulting paragraph 3 would read: States must ensure individual accountability for their use of laws in accordance with their obligations under international humanitarian law. Since accountability cannot be transferred to machines, appropriate steps may include legislative and other measures to prevent and suppress violations of IHL, such as effective domestic mechanisms for the investigation and reporting of incidents that may involve violations of IHL. In paragraph five, which we strongly support, we would recommend adding to the chapeau a reference to automation bias and to the lifecycle of the weapon system, such that the chapeau would read, "To promote human responsibility and accountability in their use of laws and to mitigate the risks of unintended engagements and automation bias. States should, when relevant to the life cycle of the weapon system, provide as the end of the chapeau. Automation bias is an important issue to address, and it occurs when commanders or operators overly trust the machines. We agree with Canada that this risk is best mitigated through personnel focused measures and belongs in box five, not box four. Additionally, we appreciate the acknowledgment by Ukraine that bias cannot be eliminated. Rather, it is a risk to be mitigated, and we recommend framing it as such. We would also move the reference to a responsible chain of human command and control, which is currently in the first sentence of box three, paragraph 6B, and in box five, paragraph four, to a new subparagraph under paragraph five that would read, "for the operation of laws within a responsible chain of command and control." This would also ensure that the concept is addressed as one of the recommended measures to address unintended engagements and promote accountability. Additionally, with this change, we believe that paragraph four could be deleted. Thank you for your time. As these were technical changes, we will provide a draft language to the Chair, and we look forward to hearing others' remarks.

Let me thank the distinguished representative of the United States for her comments, suggestions, and indeed for providing the chair with their proposals. Next on my list is the Russian Federation, to be followed by Austria. Let me first pass the floor to the distinguished representative of the Russian Federation. So.

You have the floor.

Thank you very much, distinguished chair. On box 5, we quite honestly do not have many comments. Most of this box, we're quite happy with it, and the amendments that we are planning to propose are mostly related, on one hand, the removal of certain repetitions, and on the other hand, with the deletion of any unnecessary detailization of things in this box. On paragraph 2, just like the delegation of the United States, we would like to ask for the deletion of the word "must". This paragraph is in line with earlier understandings reached in this group, and we believe that we can keep the language that was agreed earlier on the basis of consensus. Paragraph 3, we believe to replace the word "accountability" with the word "responsibility." And also to add a reference to the use of means or methods of warfare. So to say, employment of the means or methods of warfare. In paragraph five, that in many ways is a new paragraph. We would like to repeat our position of principle about the fact that any unnecessary detailization or any attempt to elaborate standard procedures or tools related to ensuring human control or human responsibility or accountability, we find such detailization to be superfluous. This is related to the fact that each of the States Parties has their national realities, national legislative systems, national procedures, and the different interaction between relevant ministries and agencies. And for that reason, to have a quite rigid spelling out of standards or procedures to ensure such human responsibility or human accountability is hardly necessary. So we would like to soften the language that we find in the chapeau of paragraph 5, and instead of the words "should provide", we would prefer to use a softer formula, "may provide inter alia on a voluntary basis". Now on subparagraph a. We would like to delete the subjective characterization that we find in the word "readily" and also at the end delete the word "and controls". Answer paragraph B. Overall, we do not have any objections against the language that we find here. Simply, we would like to delete the subjective, non-legal qualifier "proper" before the words "use of laws". Delete the word "proper". Subparagraph would prefer the complete deletion of this subparagraph because we believe that it repeats what we already have in subparagraph. Because the rules of engagement or other directives or orders related to the use of such weapon systems, all of that falls under the understanding or the word guidance that is a comprehensive notion, and it includes the different normative legal legislative documents that may relate to the use of such weapon systems or in general apply to such or to these weapon systems, including the stage of using such systems. And this is why we feel that the idea we have in subparagraph is covered already by what we have in subparagraph. Thus, subparagraph is superfluous. Paragraph 6, we would also prefer the deletion of this paragraph. We believe that this subparagraph goes a bit too far, and it already infringes on the competencies, the internal competencies of states parties. And we already said on numerous occasions that language that relate to any interference in the internal affairs of States Parties, something that we find to be inadvisable and not really relevant in the context of the objectives that are being discussed here in this group. And the identification of the specific mechanisms, tools, and procedures at the national level in connection with the regulation of any questions related to these weapons systems is purely a prerogative of the state party, and only a state party should determine what specific measures, internal mechanisms, or national procedures will be used for those different purposes. in order to ensure compliance with international humanitarian law at the national level. This is why we will repeat and stress our position of principle on this question here in this group. In the elaboration of relevant understandings, we are not in a position and we are not ready to carry out such a discussion in connection with any internal procedures, mechanisms, or tools related to ensuring compliance with obligations under IHL. We'll say it once again. It is sufficient to have understanding about the fact that states, parties have an international responsibility to ensure compliance with obligations under IHL, and also they are responsible for taking the necessary measures, procedures, and mechanisms at the national level. Now, what type of mechanism, what type of procedures will be adopted or used at the national level are not linked in any way with questions related to responsibility or accountability. On paragraph 7, once again, for the very same reason that I just referred to, we would like to replace the word "legislative" with the word "legal". and also have more clear language related to what specific violations are we talking about, not simply limiting ourselves with a reference to only IHL, but to specifically spell out that we're talking here about violations of international law, including international humanitarian law. Thank you.

I thank the distinguished representative of the Russian Federation for stating the position of his delegation and for making proposals. Next on my list is the distinguished representative of Austria. So you have the floor.

Thank you very much. Originally, we were largely fine with this box. We only had one comment when it comes to part of three, but I will come to that later. But just reacting to some of the ideas and proposals that we heard, I think in general, when it comes to accountability and responsibility, this is an area of our work where we have a lot of useful agreed language that has been also used here in the different paras, especially 2, 3 and 4. So we don't think that it's extremely useful to change the language that comes partly of 2018 report, 2019 report and the guiding principles. And just to stick what we have already, and this is to some extent is also related to our comment on para 3, but I will just go through the text maybe in general. The new addition of para 1, based on a proposal by Pakistan, is very much welcome. I think this is an extremely important addition. Para 2 is a mix of guiding principle B in the second part and a formulation out of 2018 report 23A. Uh, I think this is something that we have agreed on and should not, uh, modify and, and, and also not downplay the accountability gap that in our work was very much, uh, uh, uh, um, kind of acknowledged by the large amount of work that we have produced on this topic already. Just as a side note and also because, um, thanks are due, usually I look up those agreed languages on accountability and responsibility in the draft articles document that was, has resubmitted. Just, um, this morning, it's an excellent source for the Greek language, especially on these kind of topics. Um, on three, this is coming from the 2019 report, 17C, um, we noted that you have switched out accountabil- responsibility with accountability, as also the Russian Federation has noted. If it is responsibility, I think the current formulation would be okay, but if we switch responsibility with accountability, we should not talk about IHL alone, but about international law in general and not focus just on IHL. I think this is a quality difference that we are having here. It depends on, it correlates to some extent. Para 4 is a combination of guiding principle B and guiding principle D, so this should also be okay for most delegations. We very much value the new additions that are included in paragraph 5. I think these are important additions to our work and just explain to States more how they could actually promote human responsibility and accountability via different measures that they can take on on a national basis. So this is useful, but we are open to some tweaks or changes if they are useful. On, and last point maybe on para 7, I think the point that violations should not only apply to IHL, but all violations of international law should be relevant in this para, I think this proposal by the Russian Federation is very useful and we support it. Thank you.

Let me now pass the floor to the distinguished representative of the United Arab Emirates. You have the floor.

Thank you, Chair. I would like to share our remarks and comments on box 5, beginning with paragraph 1 thereof. We have no comments on paragraph 1, no suggested modifications or amendments to make to paragraph 1. We are of the view that legal obligations need to be duly assumed by states and individuals, not by machines. And that view is in line with IHL. This is a stance that we have voiced previously. That said, I come to paragraph 2. Here we support the principle enshrined therein, and we would recommend that we maintain the element according to which the applicable international is applicable to the entire life cycle of the weapon system, and that this needs to be part of a responsible approach to command. We would propose that we add a reference to article 1 of the Geneva Convention here in paragraph 2. Now, paragraph 3, we believe that there's an overlap, a repetition with some of the content of the previous paragraph, paragraph 2. So, in our view, it would be a good idea to merge paragraphs 2 and 3, recalling that that there must be oversight of civilian or military law in these systems. Now, on paragraph 4, this 4 is in line with our longstanding position to it that it's important that responsibility as a concept covers the entire life cycle, or accountability, I beg your pardon, covers the entire life cycle, and that there will be a need for periodic reviews to ensure that the relevant laws are being upheld. On paragraph 5, in principle, we agree with it, particularly when it comes to the references to training and the operational parameters. We would just suggest that we add wording that indicates clearly that these measures need to be linked to the national legislative and normative framework to ensure that there is effective compliance with obligations. Now, with regard to some of the subparagraphs in paragraph 5, in particular, subparagraph a, 5a, it's important that this subparagraph is more clear, more explicit. So we suggest that we add a sentence or wording there too. which says something along the lines of in case of tension or in case of combat or in instances of tension and combat to ensure that these human machine interface and controls are applicable to these conditions of tension and or conflict. With regard to 5B, we support the idea of national guidelines that are duly aligned with IHL. Now, states might have different implementation mechanisms in line with their different legal systems. And so we would recommend that the guidance or the guidelines be tailored to the command structure in each individual state, i.e. not a one-size-fits-all. On 5C, we support this idea of training on the capabilities and limitations of laws. And so we've no modification or suggestion to propose on 5C. It's fine as it is. Turning now to 5D. We support rules of engagement for laws and along those lines, we recommend that we speak in 5D of urgent systems that allow us to control or ensure oversight of laws. In paragraph 6, we call for domestic or national mechanisms for implementation and investigation. We suggest that a reference be made to the use of independent national mechanisms as a benchmark for those actions. We believe that this paragraph, although it covers a good concept, needs to be more detailed, particularly when it comes to domestic relevant or competent authorities, national mechanisms, making it clear that this might include military oversight bodies as well as independent oversight bodies. This, of course, would be tied to the vision of each individual state, but a reference to the possible of military oversight is important. Now, paragraph seven, this again is aligned with our own domestic or national commitments. ensuring that there are no violations on our domestic territory. We believe that we could fine-tune and strengthen the wording of paragraph 7, calling for the establishment of specialised implementation units for accountability and investigation in cases of violations involving laws. We think that it should be possible to merge paragraphs 2 and 3 to bolster the legal aspect and to avoid any repetition. Similarly, we think that we could merge paragraphs 6 and 7 on national additional responsibility mechanisms, or a subheading along those lines. We think that 6 and 7 fit under that concept, so they could be merged. Thank you.

I thank the distinguished representative of the United Arab Emirates. I have now two more still on my list, and that might trigger some reactions also. I have Sweden and Campaign to Stop Killer Robots. We have no time for that anymore, I'm afraid, because it's 6 o'clock and we will lose translation then, and we have to respect multilingualism. So I would propose that we continue tomorrow morning. We almost succeeded in doing a full reading, so I think we can be relatively satisfied with ourselves. And then I would like to propose that we have, again, a look at box number one. We have been doing some work, the Chair's team, and we will be sending that around. So you can find that in your e-boxes, so you have some entertainment this evening already provided by the chair, um, and we will, uh, I'm looking forward to see you again tomorrow morning, um, s- sorry. Sorry, and if you would, for one reason or the other, if you won't receive it, you can trace it yourself on the, on the, uh, on the, your, you know, the meeting place, uh, where we have all the other documents, this will also be placed there, so if you don't receive it, you can get it yourself. Um, I wish you a, a pleasant and thorough reading, um, a good evening and see you tomorrow morning at 10 o'clock, this meeting stands adjourned.