6th Meeting - 1st Session Group of Governmental Experts on Lethal Autonomous Weapons Systems 2026
Machine-readable formats: Plain text · JSON
Transcript
Transcripts available through this tool are created by using automatic speech recognition and are not official records nor official documents of the United Nations. Official records and official documents are available on the Official Document System of the United Nations. Learn more
Good afternoon, Excellencies, distinguished delegates. Could I ask you to take your seats again? Unless, of course, you are now finalizing a box for me, that's also good. Good. Welcome back. I hope you had a good lunch break. I have one, two, three, four, six more requests for the floor. Israel, Cuba, Brazil, Singapore, India, Canada. But let me start by giving the floor to the distinguished representative of Israel. You have the floor.
Good afternoon, Chair. Good afternoon, colleagues. I'm sorry to say we haven't been able to fix for you any box for the moment, but I can comment on box 4. Regarding the chapeau of box 4, while we appreciate the intent behind the new chapeau, we consider that its formulation warrants greater caution as to its systemic implications. By stating that the measures set out are in addition to legal obligations, the chapeau may unintentionally imply that practices referenced elsewhere in the text, particularly those previously included in box 3, paragraph 7, all constitute binding legal obligations rather than non-binding, policy-oriented or best practice measures. But this is not the case, as clearly some of the previous provisions do not constitute legal requirements under the existing IHL. Regarding paragraph 3, with respect to the term "context-appropriate human judgment and control," we refer to our previous comments and ask, like the US delegation, to delete this paragraph. Alternatively, should the paragraph be retained, we will suggest the following formulation: ensure there is context appropriate human judgment and control during the life cycle of laws. But this is not of our preference. Regarding paragraph four, in order to clarify the underlying point of this paragraph, we suggest including the words to its users so that it will read, ensure that the capabilities and limitations of laws in their expected use and their anticipated effects in different circumstances are understood to its users. Regarding paragraph 5 and 6, we have heard some divergent opinions on them today, and we will consider them. Thank you.
Let me thank the distinguished representative of Israel for commenting and coming up with other proposals. Next on my list is Cuba, to be followed by Brazil. But let me first pass the floor to the distinguished representative of Cuba. You have the floor, madam.
Gracias. Thank you very much, Chair. Chairman, before I turn specifically to box 4, I just wanted to reiterate an important element of Cuba's position regarding the process that we're undertaking here. The expected goal of this process, in Cuba's view, would be, at the end of the day, to adopt A legally binding international instrument which would prohibit the possession, manufacturing of totally autonomous weapons and that would establish regulations for semi-autonomous weapons. We think that this is the only way of actually meeting the goals of the convention on certain conventional weapons. And it's with this spirit that we participate in this process, hoping that this will indeed lead to specific norms which are necessary for this category of weapons. Turning to box 4, which we are currently looking at, we think that we should limit the elements that would constitute obligations of the states under the instruments that they are a party to and to those that establish voluntary mechanisms or resolutions, such as the exchange of good practice. We should separate those different elements. We can agree with the proposals made by some delegations that spoke earlier to merge paragraphs 5 and 6. And furthermore, to include in the wording the idea of strengthening measures to prevent bias. So these can be merged. Also, our delegation could agree with a proposal made by other delegations to include language on measures to prevent lethal autonomous weapons systems being transferred in some way to non-authorized operators. Thank you.
I thank the distinguished representative of Cuba for commenting on the text and making proposals. Next to my list is Brazil, to be followed by Singapore and India. But let me first pass the floor to the distinguished representative of Brazil. You have the floor.
Thank you. Thank you, Mr. President. On box 4, like others, we believe that this text is very close to what should be a desirable end product. While we remain open to any changes aimed at enhancing the clarity of the language, we would not favor the suppression of paragraphs that would fundamentally alter the substance and breadth of this box. My delegation has a few proposals and comments in reaction to our discussions yesterday and today. As to the comments made by a few delegations with regard to the formulation, "in addition to their legal obligations," in the chapeau, we would like to recall that even if Article 36 of Additional Protocol I is indeed where a positive obligation of carrying out legal reviews is spelled out, we understand that all States should assess the legality of new weapons, means and methods of warfare, regardless of whether or not they are party to Additional Protocol I. This requirement flows logically from the truism that states are prohibited from using illegal weapons, means, and methods of warfare, or from using weapons, means, and methods of warfare in an illegal manner. Like Ireland, we are open to any language solutions that do not dilute existing legal obligations and the appropriate guiding principle on this issue. As for paragraph 2, the proposal by the US delegation to replace the formulation, quote, "significant impact on target identification" and so on and so forth, end quote, with the formulation that, open quote, "that would make it a new loss," end quote, would not, in our view, resolve this issue. It would not be clear what sort of modifications to a new system would render-- to a system would render it a new system. The language contained therein is, as we have it, is much more specific and tied to the essential problems of target selection and engagement. On the other hand, we look favorably to the language proposal made by Turkey on this paragraph, paragraph 2. On paragraph 4B, we support the proposal made by Russia on the inclusion of command and control. Turning now to paragraphs five and six, my delegation believes that the issues of machine related bias and cognitive bias should be kept separate. We believe as well that both paragraphs should incorporate a reference to the need to detect unintended and automation biases. So we would propose that both paragraphs incorporate this word after prevent. On paragraph five, we also support Austria's proposal to remove as much as possible. And we believe as well that it should be advisable to bring back the reference to AI bias and insert algorithmic bias after bias in data sets. As to the US proposal to relocate Paris six to box five, we are ready to study this proposal. And with regard to the formulation of paragraph 6, we welcome the proposal made by the ICRC with a view to broadening the scope of this paragraph to include other forms of cognitive bias. Finally, we would like to express our strong support for the proposal by Belgium on the inclusion of a new paragraph on the prevention of diversion of laws to unauthorized end users. This aligns with a proposal made yesterday by India on box 3. And we are flexible as to the placement of this element, but we attach greater importance to its inclusion. I thank you.
I thank the distinguished representative of Brazil for stating the position of his delegation and commenting on the proposals of other delegations. Next on my list is Singapore, to be followed by India. But let me now pass the floor to the distinguished representative of Brazil.
Thank you, Chair, from Singapore. Good afternoon, colleagues. My delegation considers box 4 to be a good, wide-ranging list of other possible measures that address emerging technologies in the area of laws, as articulated in the mandate of the GGE. In our view, the national level measures listed in box 4 should complement the respective obligations of states under IHL. This would align with the phrase, "in addition to their legal obligations," in the chapeau. Turning from the chapeau to the second sentence of paragraph 1, we support the explicit reference to the conduct of legal reviews in box 4. Doing so would strengthen states' ability to comply with their IHR obligations where the use of laws is concerned. My delegation, therefore, welcomes the voluntary exchanges of relevant practices between states. Such voluntary exchanges would be helpful for all States in enhancing their existing institutional capacity to conduct legal reviews. These exchanges provide a good avenue to better understand how legal reviews can be conducted and tailored to specific national contexts and technological developments. On paragraph 2, we reiterate our view that the word "significant" or related synonyms such as "substantial" as suggested by some delegations this morning It is critical. Not all modifications to a loss may be material to the key functions of a loss. An example of such a modification could be the changing of the camera on the system due to obsolescence. Another example could be the enhancement of image processing software, such as those to reduce noise, improve aspect ratio for subsequent computations. My delegation's view is that it would be neither necessary nor feasible to engage in legal reviews for such modifications. In short, my delegation's view is that the modifications would have to be significant in order to trigger another legal review. And lastly, on paragraph 4, my delegation proposes minor improvements. The paragraph would now begin with "ensuring the conduct of testing and evaluation." This edit would make paragraph 4A consistent with paragraph 4B and reflect how the conduct of testing and evaluation is just as important as the provision of appropriate guidance, training, and instructions. We further suggest the addition of the phrase behind, I quote, "to the extent relevant and practicable." This would align paragraph 4A with the reality that there are inevitable limits to the ability to replicate real-world environments virtually. The revised entire paragraph would thus read: "Ensuring the conduct of testing and evaluation, such as realistic simulations of their use in operational environments, to the extent relevant and practicable." I thank you, Chair.
I thank the distinguished representative of Singapore for sharing the views of the delegation. Next on my list is India, to be followed by Canada, but let me first pass the floor to the distinguished representative of India, so you have the floor.
Thank you, Mr. Chair, and good afternoon, colleagues. Chair, we appreciate you for your steady steering of our work. We also thank you for your inclusive and consultative approach. Our view is that even though we are moving slowly, we are moving in the right direction. My delegation has an interest in strengthening the box 4. We want to advance more operational and technical safeguards to ensure that laws work as they are intended to. We view box four measures in the broader context of risk mitigation. These are different in nature from obligations which are outlined in box three, which are required to primarily clarify how IHL implementation can be ensured. This distinction is important to hold. In the context of this broad framing I provided, we would like to suggest change in the chapeau paragraph. This is very much in line with what some other delegations had to suggest in the morning. The chapeau may be amended as follows. Prior to potential deployment and as applicable throughout the entire life cycle of laws, states should inter alia consider implementing the following risk mitigation measures. The formulation, amongst other things, suggests that the underlying measures are only an illustrative list and not an exhaustive one. The term "in accordance with their legal obligations" should be used separately in paragraph 1 and paragraph 2, which are on legal reviews. This will put it in line with the legal position of several delegations on legal reviews who are not signatories to Additional Protocol. Complementary to this, my delegation is also willing to strongly encourage all States, not only those that have legal obligations on reviews under Additional Protocol, to conduct reviews, whether these are legal or even technical in nature. On paragraph two, we have taken note of the diversity of views. We do agree that terms such as significant impact are sort of ambiguous. We are not sure of the term further development or modifications pertain to hardware or software or to both. We also have questions as to whether a periodic software upgrade would be sufficient to meet the criteria of a review. We have not established a view on retaining or removing this paragraph and we will get back once there is more clarity. On paragraph 3, my delegation has heard views on the debate around conceptual idea of on control and human judgment and the need to take into account contextual aspects. We have suggested that this phrase will best act as a broad conceptual notion and not as an actionable measure, and this is best placed under box 2. We may therefore decide on its placement in this box. We support in this box measures that you proposed on testing and evaluation. on ensuring guidance and training, and the need to remove bias in data sets and automation. These, in our view, are important measures which we can support in its current formulation. In what would be a supplement to this, my delegation suggests a proposal with a similar idea, which would be to promote robust testing, evaluation, validation, and verification, which is also termed as TEVB, and encouraging states to adopt measures in their national framework to integrate such requirements. My delegation had also suggested a proposal yesterday in box 3 to address the very real risk of unauthorized actors, including terrorists and non-state armed groups, or even criminal network groups, to carry out indiscriminate attacks using laws. We thank the delegation of Belgium today, who have suggested a similar proposal on diversion. We believe that the ruling text should enshrine a measure that states adopt measures to not transfer laws to unauthorized actors, and of course, we had shared some language in that regard in box 3. We would also like to insert, if that is possible, another measure on ensuring physical security and non-physical safeguards against hacking. This would address broadly the security risks associated with loss. We have an already agreed formulation on that, which is the guiding principle F, that talks about physical security, appropriate non-physical safeguards, including cyber security against hacking or data spoofing. as well as risk of accusation by terrorist groups and the risk of proliferation. Related to this, we suggest considering another agreed understanding from the 2019 report, para 23A, which talks about various risks that pertain to laws and which need to be considered across the life cycle, including civilian casualties, incidental loss of life and injuries. It also talks about other risks, such as unintended engagements, risk of loss of control of system, et cetera. We can also consider supporting the proposal on reference to terms such as predictable and reliable, which was proposed by the delegation of Ireland. In box 4, we suggest encouraging states to wherever feasible make laws predictable and reliable. My delegation fully understands that these terms are not defined and are not legalistic in nature. They're also subjective without specific benchmarks. However, we believe that their insertion may be useful as it reflects some useful principles, consideration of which can minimize risks and reduce incidental loss of life and damage, and raising expectation around responsible use of laws. We are willing to therefore formulate it not as a binding obligation, but to emphasize its voluntary uptake to the extent these are feasible. I thank you, Mr. Chair.
I thank the distinguished representative of India for stating the position of his delegation and for making comments on the text and remarks by delegations earlier. Next on my list is Canada, to be followed by the Philippines and Norway. But let me first pass the floor to the distinguished representative of Canada. Sir, you have the floor.
Thank you, Chair. Good afternoon, colleagues. Like others, my delegation is largely supportive of Box 4, and we have listened carefully to the proposals that have been made by other delegations. We note that Box 4 is intended to address the measures required to ensure the safe and responsible development and use of laws in compliance with IHL. We recognize that the delegations in the room do not all share the same legal obligations in this regard. With that understanding, we would like to offer a few reactions to the proposals presented today. Firstly, in the chapeau, my delegation supports the proposal made by Australia to reflect the necessity for states to undertake measures consistent with and in addition to their respective legal obligations. We feel this helps give clarity on the differential obligations held by states, but nevertheless reflects the requirement to undertake the full range of measures to comply with IHL. Moving to para 1, we support the sharing of best practices between States on weapons reviews, and we believe this is an important step to address evolving practices consistent with emerging technologies. On para 2, my delegation agrees with the suggestion made by Singapore, Japan, and other delegations to replace "significant" with "substantive" as a workable alternative with regards to how impacts are characterized. The term substantive more clearly describes changes which impact system behavior rather than relying on unspecified degree of change to the internal workings of a system. On paragraph three, we believe that this paragraph is duplicative of box three, para six in respect to the concept of context appropriate human judgment and control. In our intervention made yesterday, we maintained that this concept is supportive to IHL, but not an IHL legal requirement. We had also questioned whether or not box three was the most appropriate location to introduce the concept. Upon reflecting on comments made here today, we believe that the context appropriate human judgment and control concept found in box three, paragraph six, should instead be incorporated into box four, paragraph three. We support the notion that context appropriate human judgment and control is applicable throughout the life cycle of laws, which should be retained within the same box. Moving on to paragraph four, we support the suggestion that subparagraph B should reflect the responsible chain of command and control, thus removing the specific reference to human operators. This would ensure that appropriate guidance, training and instructions are provided to all responsible personnel involved in the authorization and operation of laws. Finally, on paragraph 5 and 6, my delegation continues to support the inclusion of the concept of biases and the risks that may impact system accuracy and reliability. We note that paragraph 5 represents system algorithmic biases, whereas paragraph 6 represents the effects on human behavior. Recognizing the likely increase in reliance of algorithmic support combined with increased decision making demands, we recognize this as a distinct risk requiring specific mitigation measures. My delegation believes that this justifies the need to separate these two issues, and we will be submitting our statement to you, Chair, shortly. Thank you.
Next, I would like to thank the distinguished representative of Canada for making proposals with regards to the 326. And next, and now on my list is the Philippines, to be followed by Norway, Germany, and IHLYI. And then I have still Colombia and China. Yes. So If there is more appetite, please indicate, and we have an idea how much time we have to reserve for box 4, because my intent is also to do box 5. But now move on with the distinguished representative of the Philippines.
Thank you, Mr. Chair. We generally support the contents of box 4, but our delegation would also like to make some comments. We would like to see clarification on the use of the words "deployment" and "employment" in the chapeau of box 4 and its paragraph 1, respectively. In military terms, we understand deployment as positioning and ensuring the readiness of a weapon to be used, while employment refers to the operational or the actual use of a weapon in combat. If we are of the same understanding of the distinction between the two terms, And as the term "employment" is used in Article 36 of Additional Protocol I, we propose the use of the term "employment" in the chapeau of box 4 for consistency and greater clarity. We are also in agreement with the delegation seeking clarification and possible change on the use of the word "significant" in paragraph 2. The term "significant" can be highly subjective and may inadvertently contribute to defeating the purpose of legal weapons reviews. Our delegation also supports proposals to incorporate the requirements of predictability, reliability, traceability and explainability in paragraph 4 or in either of its sub-paragraphs. We support Ireland's proposals in this regard. We also support Belgium's suggestion to include a paragraph to deal with the risk of diversion of loss to the illicit market and/or to unauthorised end users. We are willing to work with other delegations in coming up with a language to realize this proposal. Thank you, Mr. Chair.
I thank the distinguished representative of the Philippines for commenting on the text and making proposals. Next on my list is Norway, to be followed by Germany. But let me first pass the floor to the distinguished representative of Norway.
Thank you, chair. In our first intervention, I addressed paragraph 1. I will now comment on a couple of the remaining paragraphs in box 4. In our opinion, box 4 lists a number of measures that are very useful to ensure context-appropriate human judgement and control and risk mitigation, which, by extension, helps to ensure compliance with IGL and also helps to address the special ethical considerations and concerns that are specific to laws. On paragraph 3, and in light of the intervention made by Russia this morning and others who have asked for its deletion, I would like to highlight how we consider that the paragraph adds value. We understand the point to be that enabling context-appropriate human judgment and control should be a concern at all stages of the life cycle, left of use. To say it differently, the measures should be considered at every step of the way, at the design stage, the procurement stage, the development stage, and so forth, to enable and ensure context-appropriate human judgment and control over the system and its use. And like others, we believe that paragraph 3 could usefully be merged with paragraph 8 in box 3, and we are also positively considering the Australian proposal to combine it with paragraph 4 in box 2. What's important to us is that we retain somewhere this notion that measures should be taken from beginning to end to ensure human control. And there is, of course, an argument for retaining this in box 4, since this box, as we see it, is somewhat more geared to the situation before deployment and use, whereas the regulations and prohibitions in box 3 are more focused on the later stages and the use phase. If we keep it here, we think that it could be formulated in a clearer way. The sentence, as it reads now, gives the impression that states should exercise control at all stages, which, for instance, in the design phase, does not make much sense, as there is no weapon to control. Rather, the point should be that the design process ensures that appropriate human judgement and control are possible. To make that clearer, we could propose that box 4, paragraph 3, read: "ensure that measures are taken throughout the life cycle of laws to enable context-appropriate human judgement and control". We support paragraph 2, but we could also support replacing the word "significant" with "substantive". to focus better on the changes that impact the behaviours of the system, as Hannah just explained. We believe that the Australian proposal to include information in the chapeau about who the capabilities and limitations of laws should be understood by is reasonable. and we think the proposal to move that information from subparagraph into the chapeau is good, but also Israel's proposal to add "by its users" at the end of the chapeau so that it would read "understood by its users", that also seems like a good solution to us. We agree further that we should not conflate the two distinct kinds of biases that are mentioned in paragraph 5 and 6, respectively. I am sure that it is possible to find a formulation that could combine them in one paragraph in a good way, but we frankly do not see the problem of treating them in separate paragraphs. We agree with Brazil that it could be good to add the measure to detect bias in both paragraphs, although at the same time we believe that detection is, to a certain extent, implicit in prevention and mitigation. We also support the call to re-include the example of bias in AI models in paragraph 5. As Austria pointed out, bias in datasets is not the only way that bias may enter the AI model. So this is still a useful addition, we think. And finally, we think Ireland's proposal to add a subparagraph in paragraph 4 that contains the concepts of reliability, predictability, traceability and explainability is a useful way to re-include those concepts in our text. Thank you.
I thank the distinguished representative of Norway for commenting on the text and making proposals. Next on my list is Germany. Let me pass the floor to the distinguished representative of Germany. You have the floor.
Thank you, Chair.
We see box 4 as moving in the right direction, and we've heard many suggestions earlier today to further improve the text. Regarding the chapeau of box four, we view this as an attempt to address differing obligations of states under IHL. Germany, therefore, appreciates the suggestions by delegations to comprehensively address this. We believe the wording suggested by Australia, reading, "consistent with and in addition to respective legal obligations," would properly cover this, but we also heard other suggestions that might work for us as well. Regarding box 4, para 2, suggestions made to change significant impact to substantive impact would also work for us. And then regarding the mention of biases in paras 5 and 6, we would like to reiterate what other delegations have also raised earlier. There is a distinction between biases in machines or algorithms and biases in personnel. We believe this needs to continue to be reflected in the text, but we can of course show some flexibility on the exact wording. We will also give some thought on the comments made by Finland earlier today that testing of systems to ensure safety should be approached in a multidisciplinary matter, including interdependencies between systems, other systems in the operational environment and cyber resilience. Though we do also share the assessment that it might be difficult to add additional text at this stage. Thank you.
I thank the distinguished representative of Germany for sharing the comments of her delegation and reacting to earlier proposals. Next on my list is the distinguished representative from IHLII. You have the floor.
We would like to take this opportunity to thank you for your continued efforts in advancing discussions within the framework of the GG on laws. We appreciate the constructive spirit reflected in box 4, particularly its emphasis on safeguards across the life cycles of laws. Regarding paragraph 1, we welcome the encouragement of voluntary exchanges of relevant practices among States, which can help enhance transparency and mutual understanding. At the same time, we wish to reaffirm concerns raised during the informal advisory meetings in January The effective implementation of legal reviews and life cycle safeguards requires capacity building and technical support, particularly for states with limited technical, financial, or institutional resources. Without such support, gaps in implementation could undermine compliance with international law. We also note that several safeguards in this section are currently framed as recommendations using the term should. Given the central role in ensuring compliance with international humanitarian law and preventing harm to civilians, we encourage the group to consider using stronger normative language to reflect the binding nature of these obligations. Regarding paragraph two, we suggest clarifying that further development or modification encompass not only structural or hardware changes, but also software updates, algorithmic retraining, data set modification, and remote technical support, including when provided by third party actors during armed conflicts. Such intervention may significantly affect system behavior, target selection, or legal risk assessments, and therefore should be subject to review. With respect to paragraph three, the concept of context-appropriate human judgment and control remains undefined. As noted in January, this term may be vague and open to multiple interpretations. Greater clarity is needed to ensure meaningful human involvement in decisions regarding the use of force, particularly when civilians' life and protections are at stake. Clear standards are essential to prevent uneven application or minimal oversight across different operational contexts. In paragraph 4, we welcome the emphasis on testing, evaluation, and training. We suggest clarifying that testing should include not only realistic operational simulations, but also stress testing under degraded, adversarial, or data-compromised conditions. Training should include legal, ethical, and accountability dimensions for those in the chain of command, not only technical operations of the system. Finally, we wish to reaffirm concerns raised in January regarding accountability and traceability. It would be useful to include explicit references to mechanisms such as record keeping, documentation of system parameters, and clarification of responsibilities within the chain of command. These measures would reinforce compliance with international humanitarian law and preserve meaningful human responsibility in the use of force. Thank you.
I thank the distinguished representative for sharing the views of her NGO. And next on my list is Colombia, to be followed by China. But first, let me pass the floor to the distinguished representative of Colombia. You have the floor.
Thank you, Chair. We agree with many other delegations related to how mature the text in box 4 is already. I would like to make some specific comments. First, we agree the differentiation that we should make in the chapeau between the binding and non-binding measures, and we heard many suggestions, and we are happy with many of them. What is important for us is to ensure that in this differentiation, we should not undermine existing legal commitments as established by Article 36 of AP I of the Geneva Conventions. The second comment, it is necessary for us to keep paragraph 3 Now, in terms of context-appropriate human judgment and control should be applied throughout the entire life cycle of laws, something that is very important. Now, we've heard what delegation said about the legal nature of such terminology, but for Colombia, this would be the minimum acceptable for us. in terms of ensuring that we apply IHL in connection with laws. Now, I would like to also say that this expression in box 3 relates to prohibitions and requirements related to the use of laws, the use of laws linked directly to ensuring compliance with IHL during hostilities. Now, here we rather focus on human control with a focus on application of law. Now, if we look at box four, the reference relates to the legal references related to the development and the life cycle, imposing the restrictions related to human judgment, that are not only related to employment, but to phases that are before and after that. And would like to include this language in box three, strengthens its operative nature when laws are employed. but its inclusion in box 4 emphasizes the preventive use and the structural application throughout the entire life cycle of laws. However, from a strictly legal point of view, both of these references in both boxes can exist because they relate to different stages of the process and ensure a cross-cutting approach. Nevertheless, my delegation is flexible in terms of where we include this language in box 3 or box 4, as long as we preserve the notion related to the requirement that human judgment and control is applied throughout the entire life cycle of laws, as was mentioned by other delegations, and I think Norway referred to that just recently. Now, we can also Third point, we can join other delegations that including the word "significantly" can undermine the objective regulation of such systems, something we wish to avoid. Fourth point, we can accept the proposal made by Singapore to bring in line the language of subparagraphs and. under 4, including the word "ensuring" in subparagraph A. And my final comment, my delegation could study and support the proposals that were made by Belgium yesterday to include an article that provides for not allowing any diversion of such weapons to non-authorized actors. Thank you.
I thank the distinguished representative of Colombia for sharing the views of her delegation, making proposals. Next on my list is China, to be followed by Panama and the University of Southern Denmark. But first, the distinguished representative of China.
Thank you, Chair. According to Article 36, Optional Protocol 1 of the Geneva Convention, all contracting parties should strictly abide by the review obligation in the research and development of new weapons. But the Geneva Convention has not provided any standards and procedures for the legal review. States should be allowed to formulate their appropriate mechanisms in accordance with their national conditions. And on the basis of voluntary, on voluntary basis, conduct exchange of experiences. For the specific proposals in this section, Like many colleagues present, China also believes that the current work direction is the correct one. The current text is rather mature. We only have two proposals or comments. Firstly, with regard to the chapter, Now we have the wording in addition to the legal obligation. This can be explained to meaning with the new legal obligations in addition to Article 36. China has noted that many colleagues have made similar comments or proposals. We propose to delete this part or change "in addition to" to "in accordance with". On paragraph 2, China understands and welcomes the addition of the word "significant". This is a helpful attempt to avoid repetitive legal review on some minor changes. China wish to keep this part. this qualifier. On the wording of "significant" or "substantive", we keep a flexible view. Thank you.
I thank the distinguished representative of China for sharing the views of her delegation, for making proposals and for also showing flexibility. Next on my list is Panama, to be followed by the University of Southern Denmark and Bulgaria. But let me first pass the floor to Panama. Madam, you have the floor.
Thank you very much, Chair. With respect to box 4, we agree with various delegations who have already spoken that, in general, we can support the current formulation. except for some minor comments that we'd like to make now, referring to sections where there can be improvements and greater clarity. In the chapeau of box 4, we support those delegations who have suggested that "should" ought to be replaced by "must". Also, the reference to "in addition to their legal obligations" is problematic. We know that you have tried to find a compromise solution, but we can take into consideration some of the proposals that have been made, as long as there is no watering down of the obligations under international law. as long as it doesn't give rise to a possibility for different levels of interpretation by states depending on what instruments they are a party to. There are some norms included in these legally binding instruments that are also part of international customary law. We want to ensure that the language in this chapeau is uniformly applied by all participants here. Moving on to paragraph 2, our suggestion here is that we delete the word "significant" and that for the same reasons which we mentioned yesterday in relation to box 3. If we use this qualifier "significant", we believe it introduces a lack of clarity as to the threshold. On paragraph 3, we believe this paragraph should be kept in the text. especially because it refers to human judgment and control, which is context appropriate. As our distinguished colleague from Colombia has eloquently explained just now, there are also other measures which ensure compliance with international humanitarian law. And it is important to maintain a specific reference to keeping human control and that throughout the whole lifecycle of autonomous systems in order not to create a legal vacuum, especially when it comes to attributing responsibility and accountability. Looking now at paragraphs 5 and 6, We agree with those delegations that have said that there should be a stronger link between algorithmic bias and international humanitarian law, particularly as regards its connection to the principle of distinction. In paragraph 5, particularly, We support Austria's suggestion that we delete as much as possible. We're also not clear about the use of the qualifier "unintended", given that human bias is not the only one that we're concerned with. We're also concerned about data which has been distorted or intentionally manipulated by this bias. We therefore support what has been said by the Stop Killer Robots campaign and also what the ICRC and Brazil have said in this respect. In paragraph 6, and relating to everything that's been said in paragraphs 5 and 6, we do agree that these two paragraphs deal with different issues and therefore they can exist on their own, that is, without merging them. And we would also support what the ICRC said to change some of the wording so that we are not simply referring to automation bias. And to conclude, let me say that we support the proposal made by Ireland. Now, our preference preference would be to include aspects relating to these issues, but we would prefer to see language from 2023, and we would support Belgium's proposal on including something on preventing diversion of these weapons. Thank you.
I thank the distinguished representative of Panama for stating the position of her delegation and commenting on other proposals. Next on my list is the University of Southern Denmark, to be followed by Bulgaria, Pakistan and the University of Utrecht. But first, the distinguished representative of the University of Southern Denmark. Thank.
You for giving me the floor. I'm speaking on behalf of the Auto Practices Project, a research project hosted by the Center for War Studies at the University of Southern Denmark. We would like to start by adding some brief reflections around the significance of the human dimension across the boxes and the specific phrasing context appropriate human judgment and control. As many delegations have expressed, Concern around the human dimension has been a consistent part of the work conducted by the GGE since 2017. Over time, this concern has crystallized into the current notion of context-appropriate human judgment and control that appears to be acceptable to many delegations in the room. We understand that delegations continue to hold different understandings about the human dimension. As summarized before, these include whether such a principle is already part of IHL, whether it is a means to ensure compliance with IHL, or whether it should be set as a new principle. We would like to focus on what unites these different understandings. There has long been a broad support and consensus among high contracting parties around the importance of maintaining and ensuring human judgment and control. This broad consensus in our mind justifies making the human dimension and context appropriate human judgment and control central and prominent across the text, rather than it only appearing in specific parts. We would also like to offer some more specific remarks with regard to the human dimension in paragraphs 4a and 4b. These remarks build on insights gained as part of the AutoPractices Best Practices Toolkit for policymakers, developers, and users of AI systems in the military domain, including laws. First, paragraph 4a on testing. Testing plays a particular role in ensuring a sufficient role for human actors. The process of testing, evaluation, verification, and validation, or TEVV, is a set of practices that is meant to, among others, strengthen understanding of how laws would function in specific contexts and how human users would or should respond. Military actors may use various tests, metrics, benchmarks and procedures to conduct TEVV that is informed by the anticipated and realistic context of operation and appropriate training data that relates to these contexts. As the AutoPractices project highlights, the use of precise metrics for evaluation in various scenarios strengthens the exercise of context-appropriate human judgment and control. This practice features continuously testing in conditions that match the foreseen context of use to the extent possible. So this matches exactly the phrase that we see in 4. This practice allows testers to evaluate how systems' performances relate to legal, ethical, and strategic considerations set at earlier stages of the law's lifecycle. On paragraph 4B, education, we want to add that educating humans, including but not limited to operators and commanders about laws. plays a key role in retaining sufficient levels of context-appropriate human judgment and control. Such training is not restricted to what I might call digital literacy. It should go beyond providing a basic understanding of the technical characteristics of the systems used. Training should also include education about human-machine teaming, that is, how humans interact with laws. This is particularly important because it can raise and strengthen humans' awareness about phenomena such as automation bias and other cognitive biases. As paragraph 4B highlights, a continuous training process about dynamics of human-machine interaction and teaming, as well as how they might affect human judgment and control, is required to mitigate biases and assumptions that are inherent to these dynamics. We would also like to highlight that a sufficient understanding of laws and human interactions with them would ideally be balanced in a way to avoid users' overconfidence in and overreliance on such technologies. Continuous education also aims to provide humans with the necessary information about the potential impact of using laws in different contexts, including strategic, operational, legal, ethical, and political implications that might arise. Human personnel that are trained to recognize, assess, and respond to outputs of laws while considering technical, legal, security, and other aspects have more chances of effectively using options for intervention and reflecting upon contesting and rejecting algorithmic outputs if needed. Thank you, distinguished delegates, for your attention, and thank you, Chair.
Let me thank the distinguished representative of the University of Southern Denmark for her comments and linking it to research being done. Next on my list is the distinguished representative of Bulgaria. You have the floor.
Thank you, Chair. Box 4 as a whole looks quite complete to us, as it contains important measures to ensure compliance with IHL, and we wish to react to some of the proposals we heard this morning and this afternoon. Regarding para 1 and the fact that legal reviews also happen to be a legal obligation for most of the delegations in the room, thus making should an appropriate verb in the chapeau, we find Norway's proposal for para 1 to be a good solution. On para 2, we accept the substantial instead of significant, we can accept, or Turkey's proposal Turkey's proposal to specify what we mean under the respective adjective to be put before impact. We appreciate the optimization of para 4. Furthermore, we see merit in Australia's proposal to move any responsible chain of command and human operator up to the chapeau of the para instead of in subpara B. On para 5 and 6, we prefer the current separation of the different types of biases in two distinct paras, automation bias and data bias, as biases can deteriorate the performance of the systems under consideration and present a challenge to IHL compliance. Thank you.
I thank the distinguished representative of Bulgaria for stating positions of his delegation and commenting on the text and proposals. Next on my list is the distinguished representative from Pakistan. Sir, you have the floor.
Thank you, Chair. My delegation considers box 4 to be generally in good shape and a solid basis for further work. The measures contained in this box are essentially for translating the obligations and restrictions set out in box two and three into practical steps that states must take across the lifecycle of laws for risk mitigation. After listening to the comments made by different delegates today, you offer the following comments on the shape. We note the use of "should" and share the concern expressed by several delegations and the ICRC that this formulation is insufficient for those paragraphs that reflect existing legal obligations. We would therefore support a differentiated approach: "must" for paragraphs reflecting existing obligations, while retaining appropriate flexibility elsewhere. On paragraph 2, We note the inclusion of the word significant before impact. We recognize the practical concerns behind this qualifier, that not every minor software update should trigger a full legal review. However, we would caution against setting the threshold too high. The autonomous functions of target identification, selection, and engagement are precisely the functions that determine IHL compliance. Modifications to these functions, even those that may not appear significant in isolation, can have cumulative or compounding effects on how the system performs in the field. We would therefore suggest either deleting significant as proposed by Portugal and Austria, or if it is to be retained, qualifying it with appropriate language such as potentially significant to ensure that the threshold does not inadvertently create. On paragraph five and six, I'm sorry, on paragraph three, we support the retention of context appropriate human judgment and control across the lifecycle of laws in this box. This paragraph serves a distinct and complementary function to the reference in box three. While box three addresses how human judgment and control must be exercised during the use of laws, paragraph three of box four extends this requirement to the upstream phase of design, development, testing, and acquisition. On para five and six, We welcome the clear distinction between data bias and automation bias, which, as proposed by Austria, are indeed separate challenges requiring different mitigation strategies. On data bias, we support the technological neutral formulation as the qualifier. On automation bias, we note that this is fundamentally a human factors issue. It concerns the risk that operators may over rely on systems output and fail to exercise independent judgment. This is precisely why the measures in para 4B on training and guidance are so crucial, critical, and why context appropriate human judgment and control must remain the organizing principle of this box. Finally, Chair, consistent with what we have argued in relation to box three, we believe there is merit in considering whether a reference to the principles of predictability, reliability, traceability, and explainability could also find a place in this box. These principles are not only relevant to the operational phase, but are equally critical during legal reviews, testings, and evaluation. and assessment of bias. Furthermore, we also support proposals raised by a few delegates regarding measures to prevent transfer of laws to non-state or unauthorized actors should also find some place in this box. We would therefore propose that the Chair consider including a reference to these principles in this box. Thank you, Chair.
I thank the distinguished delegate from Pakistan. Next on my list is the University of Utrecht. And at this moment, the last on my list is the United Kingdom. But let me first move now to the University of Utrecht. You have the floor.
Thank you, Chair. I intervene once again on behalf of the Asia Pacific Institute for Law and Security delegation and my own Utrecht University. We see box four as containing a range of important compliance measures, and we appreciate the efforts, Chair, of you and your team to better organize and streamline the language in this box. Our most fundamental concern with this box, also shared by others who have intervened today, is that it does not sufficiently distinguish between legal obligations and other measures. This blurring occurs as a result of the conduct of legal reviews being placed under a chapeau that use the modal verb "should." Many delegations have underlined that the obligation to undertake legal reviews of new weapons constitutes a legal obligation, at least for those States that are party to AP1. We have noted this as well and expressed concern that the inclusion of text whereby legal reviews should be conducted risks diluting this obligation. In this version of the rolling text, the issue has become exacerbated by the introduction of the phrase in addition to their legal obligations in the chapeau and by the removal of the phrase in accordance with their obligations under international law from paragraph one. Chair, we recognize that different states are in a different legal position when it comes to the obligation to conduct legal reviews. The language of the rolling text must therefore be sensitive to differing legal obligations, but not simply adopt the lowest common denominator. We note that the group's earlier consensus documents have managed to resolve this problem. For example, paragraph 23 of the 2023 report noted that, and I quote, "In accordance with states' obligations under international law, in the study, development, acquisition, or adoption of a new weapon, means, or method of warfare, determination must be made whether its employment would, in some or all circumstances, be prohibited by international law." This formulation accommodates both those states that are under an obligation to conduct legal reviews and those that are not. Furthermore, paragraph two of this box concerns modification of existing laws by requiring a renewed review to ensure compliance with applicable law. While Article 36 of AP1 does not explicitly use the term "modification," it is generally understood in state practice that modifications that may affect a weapon's characteristics or effects trigger the need for a new or updated legal review. Accordingly, paragraph two also reflects an aspect of an existing obligation for those states that have this obligation, rather than a supplementary commitment. For these reasons, prefacing paragraphs one and two with the clause with, in addition to their legal obligations, is inconsistent with existing law and must be avoided. Additionally, as others have mentioned, we recommend removing the term significant as it may narrow the scope of the paragraph. We believe this term lacks the clarity as to the threshold, much in line with the discussion on paragraph 7C in box 3 yesterday. Chair, we recommend that the substance of paragraphs 1 and 2 be moved before the current chapeau and reordering the language for a better flow. We further recommend that these paragraphs draw on the consensus language of the 2023 report while substituting a new laws for more generic terms, a new weapon, means or method of warfare that is used in 2023. Thus, we propose the following three paragraphs, all to be placed at the beginning of box 4, before the current chapeau. I will read them out slowly and provide the text to the Secretariat. In accordance with States' obligations under international law in the study, development, acquisition or adoption of new laws, a determination must be made whether its employment would, in some or all circumstances, be prohibited by international law. This also applies to any further development of, or modifications to, existing laws that have a significant impact on their target identification, selection, and engagement functions, or on their anticipated effects. In this context, the voluntary exchange of relevant best practices between states is encouraged, bearing in mind national security considerations or commercial restrictions on proprietary information. Chair, in subparagraph 4A, which was previously paragraph 5, the removal of the word "rigorous" creates a materially weaker compliance measure. The word "rigorous" signals a certain qualitative threshold and implies that testing should be methodologically sound, systematic, sufficiently extensive, stress tested under varied conditions, and so on. Without that qualifier, the clause merely requires that some testing and evaluation occur, potentially transforming the testing requirement into a mere procedural formality. In the same subparagraph, we suggest retaining the clause that such testing is to enable a human user to have a reliable expectation of how laws will perform in the anticipated circumstances of its use, as it explicitly connects testing to an outcome. It is rigorous testing that makes a reliable expectation possible, not just any testing. Moreover, this proposal is consistent with the suggestion of multiple delegations to reflect the notions of predictability and reliability in the rolling text. In paragraphs five and six, a reference to the detection of bias has been lost. Removing detect eliminates an explicit compliance measure to monitor and audit systems for bias. We should look at detect as an activity aiming at actively identifying bias through testing and auditing, correct as an activity actively remedying identified flaws, and mitigate as an activity directed at reducing residual risk. The revised version significantly weakens the commitment to this three-part dynamic lifecycle oversight of monitor, correct, and mitigate. I thank you, Chair.
Let me thank the distinguished representative of Utrecht University for commenting on the text and making proposals. Next and last on my list is the United Kingdom. You have the floor.
Thank you, Mr. Chair. The UK appreciates this continued fruitful discussion on the contents of box four, and we have taken time and are continuing to take time to consider the wide variety of comments and suggestions that have come from delegations today. Moving on to the text itself and moving through it paragraph by paragraph, on the chapeau, the UK can support the suggested change to the reference to legal obligations that was proposed earlier by Australia. With reference to the term in the chapeau, prior to potential deployment, We think the term deployment here is open to interpretation. For instance, would this refer to a system being introduced into service or one that is being physically moved to or placed in a theatre of operation? Therefore, we propose amending prior to potential deployment to prior to introduction into service. Moving to paragraph one, for the second sentence in paragraph one, We suggest an addition to add some clarity. The proposal we have is would be for the sentence to read: "In this context, the voluntary exchange of relevant practices related to the conduct of legal reviews between States is encouraged, bearing in mind national security considerations or commercial restrictions on proprietary information." Paragraph 2. The UK recognises the difficulties of the use of adjectives within the rolling text. The term significant impact here is not one that is recognised, and therefore it introduces another term that would need to be aligned with terms used in practice. As we said this morning, we would suggest that the term "material" is more in keeping with practice, and we agree with the number of interventions on this box who have made the same point. Paragraph 3. In conjunction with our comments on context-appropriate human judgment and control in box 3, the UK suggests the following amendment to this paragraph, which would now read: Ensure context appropriate human judgment and control is applied across the life cycle of laws in a manner that ensures they can be used in compliance with IHL and that accountability is maintained for their effects. Moving on to paragraph four, we have some concerns around the term realistic simulations of their use in an operational environment. We believe that what this means is subjective and not being able to achieve a realistic simulation may not mean that a system has not been suitably tested for the purposes of being legal under IHL. Moreover, the term realistic simulation refers to a type of technology. The UK is concerned that making specifications around the technologies used to test laws makes unhelpful assumptions around current and potential future testing. Removing the reference to realistic simulation will be consistent with our approach across the rolling text, where we have removed terms such as machine learning in order to reduce the risk of the text becoming redundant due to technological change. Furthermore, it would be helpful here to add text that focuses on what we believe we are trying to achieve with this paragraph. i.e. that testing and evaluation is to provide evidence and confidence that the laws can be used in compliance with IHL. Therefore, we offer the following amendment to para. 4, namely: "conducting testing and evaluation, including defined testing, evaluation, validation and verification frameworks that are to be used throughout the life cycle of the laws and provide verifiable evidence on the expected performance of the laws. Moving on to paragraph 4, the UK suggested an amendment to include those who are responsible for testing, training and maintaining laws. Our proposed amendment would change that paragraph to read: "ensuring appropriate guidance, training and instructions for those in the responsible chain of command and control human operators of laws, and those responsible for maintaining, testing, training and otherwise modifying laws. In addition to this, we note that there is no current attention being given to those who may carry out an act that could lead to the need for a legal re-review. Therefore, we suggest the following additional sub-bullet, which would read: Ensure that those maintaining, testing, training or otherwise modifying laws are provided with suitable guidance and training to identify when their work would result in a requirement for a further legal review. Paragraphs 5 and 6, we wonder if we need the term "as much as possible", and therefore for both paragraphs we would suggest amending it to the term "as far as reasonably practicable". Specifically, the UK delegation thinks that there are two key approaches to addressing prevention and mitigation. At least in our understanding of the term, we cannot completely remove bias. We would also like to append some text to the paragraph to include a reference to algorithmic bias. Therefore, we suggest that the paragraph should read: "Implement measures to reduce bias, including biases in data sets and algorithmic biases, and mitigate residual risks presented by remaining bias or biases, in the plural, as much as is practicable. Paragraph six, similar to the comment I've just made, we would like the paragraph to say prevent or mitigate. And in these paragraphs, in response to the suggestion from our distinguished Russian colleague to revert to previous texts which dealt with the different concepts of bias together, so referring to the concepts in one place, the UK recognizes that whilst both paragraphs refer to bias, we want to make the point that unintended bias and bias in data sets and automation bias are two very different concepts and require different measures to address them. We therefore believe the paragraphs are best kept separate as they currently appear. Thank you, Mr. Chair.
I thank the distinguished representative of the United Kingdom for sharing the views of his delegation and a very thorough common document will be produced. I have one more look around the room. Now, there are no further requests for the floor. Then, just looking back for a moment at our discussion. Well, I think we quite succeeded in making you collectively unhappy with what we did with regard to legal obligations. Clearly, we have homework to do. We have a number of options to consider. That will be a serious job, but we will take that up, of course. We have had different views, again, of course, with regard to context-appropriate human judgment and control, whether to have it here, and if to have it here, how to formulate it. That is part of a larger discussion also that was triggered already in box 3, para 6, if I recall correctly. Adjectives remains a difficult issue. Many proposals also for fixes. We will look into that. I note also the proposal by Belgium. We will look into that. Thank you for sending us the proposal and also others for reflecting on that. I think Yeah, and then on bias, we had already a long discussion whether to have that in different powers or not. I acknowledge the proposal also done by one delegation on that. We have to look into that. I think that's it for the moment. And thank you also in advance already for sending in your proposal so we can have a close look at it. when we go through the text with regard to the second reading later this week. If you just give me a moment, I will take my speaking notes with regard to box number 5. I'm a bit hesitant to read out what I have here, but I still do it. That is, this box has received generally positive feedback, but that will not exclude, of course, having more feedback now on what we produce. So let's see. For my part and from the side of my team, we have modified some parts of it. That was mostly concentrated regarding, focused on removing redundancies with other parts of the rolling text. Let me explain just a few things we did. In power two, we combined the form of powers three and four in line with comments we have heard on there being a certain level of duplication. Para 3, which used to be para 5, we moved it up, and the sub-paras were retained, at the same time also streamlining the text to cut parts that were duplicative. That's it, what I have from my side as a kind of a -- sorry. As further information with regard, uh, to what we have changed in, in the text, let me now open the floor for your comments on box number five. So far, I've one request for the floor, so let's see whether that triggers further request for the floor. And I see still people also being busy with the paper, so let me first pass the floor to the distinguished representative of Bangladesh. You have the floor.
Mr. Chair, thank you for giving me the floor. The rolling text refers to individual accountability in box 5. However, it does not clearly reaffirm the principle of individual criminal responsibility. Last year, my delegation explained in detail why this principle must be explicitly recognized in the regulation of laws. We renew that call today for justifiable reasons. Mr. Chair, my delegation reiterates that laws threaten justice and accountability in armed conflict. International criminal justice is built on holding human beings responsible for their decisions. When lethal decisions are delegated to machines, the link between human judgment and legal responsibility under IHL and ICL is broken. This makes it extremely difficult to attribute responsibility for grave violations. Responsibility cannot be outsourced to algorithms. Criminal law applies to humans, not machines. If no person can be held accountable for unlawful killings, impunity will prevail. The international community has worked hard to end impunity by enforcing individual criminal responsibility for international crimes. Yet, the autonomous nature of laws risks shielding individuals from liability. It undermines the core principles of responsibility and attribution. This concern becomes even more acute in light of an alarming emerging trend. The commission of genocide and crimes against humanity may be increasingly incentivised if greater advantages are attached to disguising mass atrocities as lawful acts of war or as collateral damage. Armed conflicts themselves may even be manufactured or manipulated to shield criminal conduct. ICRC report indicates that armed conflicts have roughly doubled over the last 15 years, with over 130 currently active, marking a historic high since World War II. If the international criminal justice system, including IHL and ICL, is weakened, it could become the Achilles heel of the international legal order. enabling grave crimes, including the crime of aggression, and undermining both justice and peace. Mr. Chair, the ruling text must therefore acknowledge this diametrically opposite relationship between international criminal justice and laws and address this clear tension. To address it, the regulation of laws requires a renewed and explicit commitment to the principle of individual criminal responsibility. My delegation repeats the following formulation: Humans, not machines, bear individual criminal responsibility for grave violations of IHL. I repeat, humans, not machines, bear individual criminal responsibility for grave violations of IHL." This sentence could suitably be placed in box 2 or in box 5, paragraph 2, after the phrase, "since responsibility and accountability cannot be transferred to machines." Mr. Chair, this principle is firmly grounded in international law. The Geneva Conventions and Additional Protocol I have elaborated it. Rule 151 of the ICRC's study on customary international humanitarian law confirms that individual criminal responsibility for war crimes is a customary law rule applicable in both international and non-international armed conflicts. International criminal tribunals have consistently upheld this principle. My delegation urges this august body to clearly reflect this fundamental rule of international criminal justice. As laws directly challenge it, the strongest response is to reaffirm it. I thank you, Mr. Chair.
I thank the distinguished representative of Bangladesh for stating the position of his delegation and for making a concrete proposal. Could you please Send it again. Send it to us. Would be helpful. Thank you very much. Next on my list is Partners for Transparency, to be followed by Australia, Pakistan and Finland. But first, Partners for Transparency, you have the floor.
Thank you, Mr. Chair.
Partners for Transparency emphasizes that promoting accountability for violations related to autonomous weapons requires users to have a practical understanding of how the weapon systems operate. Therefore, states should review weapons as a positive means to enhance accountability and information sharing based on Article.
36 of the Additional Protocol I of the Geneva Conventions of 1949.
These reviews necessitate complex procedures to test weapons, performance, and assess the risk associated with unintended loss of control. Consequently, Partners encourages states to amend the following paragraphs. In paragraph 3, point A, we suggest adding the phrase, "allowing the user a practical understanding of how the system operates, specifically the technical triggers that lead to the use of force." In paragraph 3 of point B, we propose.
Adding the phrase, "stressing the importance of information exchange and best practices among states regarding these policies, principles, and procedures.
To ensure the highest standards of compliance and transparency.".
Paragraph 3.c suggests adding the phrase "with adequate awareness.
Of the operational context of deploying the system, including surrounding civilian facilities, types of targets, geographical areas, and the specified.
Duration of the operation.".
In paragraph 5, we propose adding a section that reads, "Given the significant technical and financial resources required for these assessments, international cooperation and information sharing concerning weapon reviews should be enhanced to identify best practices.
And reduce costs associated with testing and evaluation procedures to mitigate the risk of unintended loss of control." Thank you.
I thank the distinguished representative from Partners for Transparency and now pass the floor to the distinguished representative of Australia to be followed by Pakistan, but first Australia.
Thank you, Chair.
Australia appreciates the refinements made to Box 5.
We will focus our intervention on one key paragraph that we consider requires amendment.
In our view, paragraph 4 should be confined to IHL. IHL is the overarching body of law governing this text and requires the prevention and suppression of violations. The paragraph should also refer to the use of laws, as IHL governs use.
Such a focus also aligns with the operational accountability focus of paragraphs 2, 3 and 5 of this box.
The paragraph would therefore read: States must take all appropriate steps to prevent and suppress violations of IHL relating to the use of laws by persons or on territory under their jurisdiction. There are also some minor grammatical improvements we believe should be made to paragraphs 1 and 5, which we will submit to you in writing. I thank you.
I thank the distinguished representative of Australia. for making some concrete proposals, and we would like to see indeed the grammatical proposals. Next on my list is Pakistan, to be followed by Finland and Sweden, but first the distinguished representative of Pakistan. You have the floor.
Thank you, Chair. We also agree that the recent refinements of Box 5 have brought us closer to convergence. We would like to share comments in three areas. two minor edits that we would like to see in this box and one, our strong support for one of the principles that already is in this box. Chair, first, on the framework and the ideas of state responsibility, especially the framing that is contained in paragraph one. It correctly places states alongside parties to armed conflict and individuals as bearing responsibility and accountability for the decisions they take with regard to lethal autonomous weapon system. It's an important principle. However, we note that what paragraph one captures is the decisional dimension of the responsibility, the obligation to exercise responsibility when making decisions regarding laws across the life cycle. What is not yet reflected in this box is the consequential dimension, namely that every internationally wrongful act of a state involving laws entails the international responsibility of that state under international law, including the attendant obligations to cease the wrongful conduct and to make full reparation for the injury caused. We believe this is not a novel proposition. It is a well established principle in customary international law reflected in the draft articles on straight responsibility. And most importantly, it is a principle that our group has already expressly agreed in paragraph 19 of the 2022 GGE report. Secondly, Chair, we believe that this is also perhaps the box where we can aim to integrate the principle of traceability within the chain of human command and control. We note that, for instance, paragraph 2 in this box refers to the operations of a system within a responsible chain of human command and control. And my delegation has previously explained why the qualified traceable is essential and not superfluous in the context of laws. Traditional command structures are inherently traceable. The chain of orders is documented, attributable, and verifiable. But in case of laws, we cannot say that in every case. In this case, traceability should mean proactively designing data logs, reporting mechanisms, and audit trails that can verify who issued what instructions, what parameters were set, and how the system interpreted and executed those instructions through its algorithms. So, indeed, if we accept the consequential dimension of state responsibility that I have just described in para. 1, then we believe traceability also becomes even more critical. because establishing whether an internationally wrongful act has occurred and attributing that act requires precisely the kind of verifiable record that traceability would provide. Finally, we would like to extend our support for paragraph 5 on internal investigation mechanisms. We support the retention of that paragraph. Effective international mechanisms for investigation, reporting, and appropriate action are a necessary corollary of the responsibility framework. And if we reaffirm that the state bears both decisional and consequential responsibility, then we must also affirm the mechanisms through which states would give domestic effect to those obligations. This is entirely within the mandate of our group and consistent with the approach that we have taken throughout the rolling text and is also in analogous provisions of other CCW instruments. I thank you, Chair.
I thank the distinguished representative of Pakistan for making proposals with regard to this. The box next on my list is Finland, to be followed by Sweden. But let me first pass the floor to the distinguished representative of Finland. So you have the floor.
Thank you, Chair. We note further positive developments in the text. Regarding paragraph 2, as before, we would have wished to see an explicit reference to a traceable and responsible chain of human command and control. Ensuring accountability requires that it must be possible to identify and trace those responsible for decisions across the entire process. Additionally, we believe that box five should contain a reference to voluntary exchange of best practices between states. This could be done, for example, by adding a sub-paragraph 3D with the following content. On a voluntary basis and with full respect for national security considerations and the protection of sensitive appropriatory information, States are encouraged to exchange lessons learned and best practices related to the development, use and governance of laws. We are open to other formulations as well, of course. I thank you.
And I thank the distinguished representative of Finland for commenting on the text and making concrete proposals. Next on my list is Sweden, to be followed by IHL, YI and the European Union. But first, the distinguished representative of Sweden.
Thank you, Chair. I'll start with some positive feedback that Sweden supports the substance of this box, but we wish to make some suggestions, mainly in order to bring the language more in line with other parts of the rolling text and existing treaties. Starting with para 1. In the last sentence of para 1, we suggest to replace "weapon system" with "laws" and "entire" could be left out. The sentence would then read, "This should be considered across the life cycle of laws." In para. 2, we suggest to remove "the means or methods of warfare involving" and add "international law, in particular, before IHL." "States must ensure individual accountability for the use of laws in accordance with their obligations under international law, in particular IHL, including through the operation of such systems within a responsible chain of human command and control. Para 3 include important measures, not merely to promote responsibility and accountability, but also to ensure that laws are used in compliance with IHL. Hence, we don't see box five to be the right place for these measures, which is why we have suggested when discussing box three that para 3A should be moved to para seven in box three. We believe that para 3B and C are captured by existing formulations in box four. In paragraph 4, "steps" should be changed to "measures", and in paragraph 5, we would like to see "must" be replaced by "should". Thank you, Chair.
I thank the distinguished representative of Sweden for making concrete suggestions to the text. Next on my list is IHLY, to be followed by the European Union and Portugal. You have the floor.
Thank you, Chair. I know all of the delegations today tried to find solutions about the accountability gaps. And actually, we stand with them under the section five, highlighting again, like what we share in January about the important to work together about the, in the ruling text, especially in section five, about the accountability gap. And to put the priority for us about the implementation of IHL and human rights. Actually, again, we back as I call for responsibility and think about responsibility to add the section about the for the private sector. Not all of you know today the private sectors play a big role in the conflict today. So we should put our priority to add some section and especially in five part about the private sector and AI provider. especially in responsibility across the full life cycle of laws. And also to think that also and link that between that about the protection of civilians and maybe also the data collection that targets the tracking and also in autonomous system. So that's, we think that is very important to listen for that. And the last point I'm sharing today about the voices of youth in Middle East and North Africa. Today, we face a lot of challenges in our region, and we need to take your attention to have a legal binding instrument. We are not part of this war, and we not needed to our region to be continue in this war. And we need real solution for that. We know as I tell why I, you make a lot of effort to have this document, and we are happy about that. but we need to think about that we should work together to have legal binding instrument. Our protection as a youth now is between your hand and we need your listen to our voices. Thank you, Chair.
I thank the distinguished representative from IHLI for her comments. and suggestions. Next on my list is the European Union, and then I have Portugal, the US, Stop Killer Robots, and the ICRC. But let me first pass the floor now to the distinguished representative of the European Union. Madam, you have the floor.
Thank you, Mr. Chair. On box 5 and human responsibility and accountability, European Union supports reference to states, parties to armed conflict, and individuals remaining at all times responsible and accountable in accordance with applicable international law for their decisions with regard to laws, since responsibility and accountability cannot be transferred to machines, and that this should be considered across the entire life cycle of laws. EU reiterates that human beings must remain accountable for decisions over the use of force in order to ensure compliance with international law, in particular IHL, and relevant ethical considerations. We are also of the view that, in accordance with states' international obligations and commitments, states should consider adopting and updating appropriate national measures to ensure effective internal mechanism for the investigation, reporting and appropriate action in response to incidents with the use of lethal autonomous weapons systems that may involve violations of international humanitarian law. I thank you.
I thank the distinguished representative of the European Union for stating the position of the Union. Next on my list is Portugal, to be followed by the U.S. Let me now pass the floor to the distinguished representative of Portugal.
Thank you, Chair. First and foremost, I'd just like to say that we are generally in agreement with the text as is. I would just like to suggest some refinement. So, first on paragraph 2, We note that, across the text, the terms "responsibility" and "accountability" are used sometimes together, although they are not necessarily complementary expressions or notions or concepts, if you will. In paragraph 2, we understand individual accountability as entailing as well individual criminal responsibility. We would therefore suggest also for the consistency of the text, suggest to write the following text, "States must ensure individual responsibility and accountability." So adding responsibility before accountability. Also, we would, on paragraph 2, in what concerns the expression within a responsible chain of human command and control, we would suggest to add the traceability element by including the following text, and I'll read, "within a responsible and traceable chain of human command and control," so adding the words traceable after responsible and before chain. Then on paragraph three, here we have the just at the beginning, the expression human responsibility and accountability. So again, and for the reasons that I stated before and just for for for reasons of consistency across the box, We would suggest to just write to promote individual responsibility and accountability, or if the colleagues prefer to retain the element human, to write to promote human individual responsibility and accountability. If we use here human individual responsibility and accountability, then of course, we would have also to ensure that the same element, the human individual responsibility and accountability, is used consistently also in other paragraphs, namely paragraph 2. In what concerns paragraph 4, we would just like to state that we agree with the reference to international law, including IHL, as is in the text. First of all, because of course, it's not just IHL that applies here, although it's lex specialis, of course, for armed conflict. And of course, CCW is within an IHL framework. But nevertheless, we cannot forget that there are other branches of international law that also apply. including human rights law and others, and even box 5 also implies in paragraphs 1 and 2 other branches such as the law and state responsibility and also international criminal law. So this is just to support the inclusion and maintenance of the expression international law, including IHL. Finally, in what concerns paragraph 5, we would like to suggest that in the expression "investigations, reporting and appropriate action", we add also "prosecution", meaning this would read "investigation, reporting, prosecution and other appropriate action". This is just as a matter of also consistency since we refer to individual responsibility and accountability across the other paragraphs in the box. So this is a matter of consistency and also to reinforce the very important element of prosecution in what concerns the respect for IHL. Finally, I would just like to say that we concur with the suggestion just made by Sweden to change the word must for shall. Thank you so much.
I think Sweden said should, but yes, so should or shall, it makes a difference.
I would suggest the shall, but could leave with should.
Next on my list is United States, to be followed by Stock Killer Robots and the ICRC, but first the US. So you have the floor.
Thank you for giving me the floor, Mr. Chair. As we've stated before, we see the value of box five to be that it articulates measures specific to laws that strengthen accountability for their use. Our proposed changes to the language in the rolling text are primarily focused on refining the existing points. In reference to paragraph one, we understand responsibility as referring to legal responsibility. and accountability as the mechanisms and efforts to establish and implement that legal responsibility. Therefore, it would make sense to refer to responsibility in the first part of the sentence and accountability in the second part of the sentence, rather than both responsibility and accountability both times. We also propose to remove the last sentence of the paragraph, which reads, "This should be considered across the entire life cycle of the weapon system." or to further clarify what is intended. As currently drafted, it is too vague to be useful. Regarding paragraph two, we propose to remove the final phrase and end the paragraph after "with their obligations under IHL." The operation of such systems within a responsible chain of human command and control is not a prerequisite for ensuring individual accountability. This language is more suitable as a subpoint in the next paragraph alongside other measures. We strongly support paragraph three and its subparagraphs, but we'll propose three refinements. The first, as just mentioned, is to add a subparagraph recommending that states should provide the operation of such systems within a responsible chain of command and control. The second refinement is to incorporate the concept of automation bias in paragraph three from its current location in box four. Automation bias means when personnel overtrust the machine. It relates to the responsibility and accountability issues and therefore makes sense to include here because the listed measures can mitigate the risks of automation bias. The revised paragraph would then read, to promote human responsibility and accountability in the use of laws, and to mitigate the risk of automation bias and unintended engagements, states should provide. We heard ICRC's comments on box four on expanding the reference to cognitive biases more broadly. We appreciate that suggestion and would be open to discussing it further. Finally, in paragraph 3C, We find the phrase "to understand" unnecessary and unhelpful, so we propose to replace "to understand" with "on." I will note that, as mentioned in our previous intervention, this subparagraph and subparagraph B are where the measures on guidance and training should sit, not in box four. Paragraph four is not a correct statement of existing IHL obligations. More of a state should not adopt this rule because states are not obliged to prevent and suppress violations by other states or parties to a conflict. We cannot join such a statement, and we would delete the paragraph in its entirety. Paragraph five also mistakes international law by characterizing a good implementation practice as an obligation. For this reason, we agree with Sweden that the word must in the paragraph should be changed to should. We also recommend changing "incidents with the use of laws" to "incidents involving their use of laws" to make it clear that each state's internal mechanism is focused on its own actions. States should not be ensuring internal mechanisms that are applicable to the conduct of other states. Finally, we echo Finland's call for placing language on voluntary exchanges of information in this box, expanding on the notion in box four, paragraph one. Such exchanges would help promote common understandings in an area and where states practices dynamic, evolving alongside rapid technological developments. It would also serve as a confidence building and transparency measure and directly inform and support future work to develop an instrument and possible technical annexes. Our proposal will be a new paragraph that could read, "States should conduct voluntary exchanges of information in a collaborative spirit about their own practice related to emerging technologies in the area of laws, including but not limited to military applications, domestic legal reviews, and other measures to implement applicable rules of IHL. and using emerging technologies in the area of laws to enhance IHL implementation supplementary to legal requirements. Thank you for your time, Mr. Chair.
Let me thank the distinguished representative of the US for commenting on the present text proposal and proposals by others and stating the positions of her delegation. I have three more requests for the floor. Stop Killer Robots, the ICRC and Australia. But let me first pass the floor to the distinguished representative of Stop Killer Robots. You have the floor.
Thank you, Chair. Stop Killer Robots welcomes that the text continues to recognize that responsibility and accountability cannot be transferred to machines stated in paragraph one. Regarding the reference in paragraph two to a chain of human command and control, we would like to emphasize that chains of command and control will and must always be human. We would like to reiterate that meaningful responsibility and accountability can only be possible if there is meaningful human control over weapon systems. Alongside elements and other boxes of the rolling text that we have already commented on, we therefore welcome paragraph 3 of box 5. This sets out further measures that can be taken to contribute to users' functional understanding of systems, which is a key component of meaningful human control. We urge States to retain all elements in the current rolling text that relate to ensuring meaningful human control and use these as a basis for the negotiation of robust and clear rules and requirements on autonomous weapons systems as part of a legally binding instrument. Thank you, Chair.
I thank the distinguished representative of Stop Killer Robots. We now pass on to the distinguished representative of the ICRC, to be followed by Austria and Switzerland. But first, the ICRC.
Thank you, Mr. Chair. Like others who have spoken, we are quite happy with the way this Box 5 has turned out, thanks to your efforts. We have a couple of specific suggestions for refinements and then just following from what the distinguished delegate of the United States just added, I have a couple of points of clarification. So with regard to paragraph 2 of box 5, we'd recommend replacing the phrase, "the means or methods of warfare involving" with the phrase, "humans who plan, decide upon, and execute attacks or other operations involving." This is to emphasize that legal obligations under IHL rules must be fulfilled by those persons who plan, decide on, and carry out military operations. So the whole sentence would read, "States must ensure individual accountability for humans," or you could say human beings or personnel, if you prefer, for humans who plan, decide upon, and execute attacks or other operations involving the use of laws in accordance with their obligations under IHL, including through the operation of such systems within a responsible chain of human command and control. The second point relates to paragraph 5. As we've already recognized in the GGE's Guiding Principle D, accountability for it's not just accountability for using any laws, it's also for developing and deploying laws. So accountability for developing, deploying, and using must be ensured in accordance with applicable international law. So because of that, we'd recommend referring to both development and use in paragraph five, and to facilitate that also to remove the reference to incidents. So we'd suggest paragraph five read, "States must ensure effective internal mechanisms for the investigation, reporting, and appropriate action in response to development or use of laws that may involve violations of IHL." Just two points in response to the intervention from the distinguished delegate of the United States. The suggestion to move the point on mitigating automation bias to paragraph three here, that seems to make a good deal of sense. We'll look at it further, but we certainly follow the logic of that suggestion. The query was on paragraph four, which the United States suggested be deleted. I'm not sure we've quite followed the reasoning, because it states the end of that sentence makes it clear it relates to persons or on territory under the jurisdiction of a state. So it's not sort of taking appropriate steps to prevent and suppress violations of international law everywhere or by other states. It's only by persons or on territory under their own jurisdiction. So I don't know if the delegate of the US might explain a little bit more, because for us this paragraph seems perfectly okay. Thank you very much, Mr. Chair.
I thank the distinguished representative of the ICRC for suggestions and reactions to previous proposals. Next on my list is Austria, to be followed by Switzerland. But let me first pass the floor to the distinguished representative of Austria. You have the floor.
Thank you very much. One point that we previously made on the discussion of this part is that we, and which is important to know, is that we have a rich set of previous agreements from different reports and from the guiding principles on accountability and responsibility, which maybe does not make our task easier because I think some of the issues that we hear today are stemming from the combination of different parts of previous agreed language into something new and also switching out responsibility and accountability to some extent. At the same time, we should also not put into question what we have agreed already too much. Just to say that, for example, in para one, I mean, much of this is from guiding principle B, which also caters for both the responsibility and accountability. I think what you have done here by combining them in a rather shorter sentence works. So we don't completely agree with the US proposals to separate them and put them into different parts. I mean, just to read out Guiding Principle B, which says, "Human responsibility for decisions on the use of weapon systems must be retained. since accountability cannot be transferred to machines, so this cannot be transferred to machines, is referring to the accountability and responsibility. And also the second sentence that was put into question is part of guiding principle B, which is this should be considered across the entire life cycle of the weapons system. So this is something that we have already agreed. When it comes to para 2, I think we would agree with the improvements by the ICRC. And also I have another point that we wanted to make, because we are talking now here only about IHL, but the original source of this text, which is Guiding Principle D, also talked about international law as a whole, and we would like to maintain this notion so that we change IHL to international law. We had some internal discussion if this then works together with the means and methods of warfare. We also had some different opinions in our delegations about this. But with the suggestion by the ICRC, this problem would be resolved and we don't have to discuss it any further. So I think this is a very good solution for this problem, and it would also allow us to include international law in this para. which also caters to the points that have been made today about the importance of international criminal law. Having said this, we would also be interested in having more language on this overall in this section as it is an important part. Moving further, we could also go along with having the issue of automation bias added to this segment as it makes sense, but we are flexible on this. It can also mirror the other paragraphs in the previous boxes. I mean, this is nothing that is a big problem for us. We are very much attached to paragraph 4. I mean, this should be maintained. We would be against deletion. I think it's an -- we think it's an important element. We would even go back to some previous formulations that have been deleted and which seem important to us, which is including States must take all appropriate steps, including legislative and other measures. Since it says including, I mean, it doesn't say that this is the end of everything, but it at least indicates that one, these measures that would implement those steps to prevent and suppress violations could also include legislative measures, legal measures, and this is important element, especially since we are talking here about violations of international law and not just some random criminal acts. So this makes a lot of sense and something to be taken serious. And in the arguments why this has to be maintained, we agree with the ICRC. We also would like to keep it as international law and not just as IHL. And we are also not completely convinced that IHL governs only the use. of weapon systems. Going further, we think that the proposal by the US to have a new para on cooperation and information exchange is very valuable, and we recall that Switzerland has made a similar proposal already in the past. So having something like this added in box 5 would make it definitely more rich and useful. Thank you.
Let me also thank the distinguished representative of Austria for commenting on the previous proposal and making suggestions. Next on my list is Switzerland, to be followed by Israel and the Russian Federation. But let me first pass the floor to the distinguished representative of Switzerland. You have the floor.
Thank you, Mr. Chair. Overall, we think box 5 is in very good shape, and we could support it as it stands. It contains important principles, especially that responsibility and accountability cannot be transferred to machines and hence needs to remain with states, parties to armed conflict and individual humans.
Building on that principle, we welcome that.
Box five sets out clear statements that states must ensure individual accountability and to undertake measures to promote human responsibility.
This is important for us as.
A key concern that sparked the very debate on lethal autonomous weapons systems was the risk of a responsibility gap. Doing so together with other measures to foster implementation and compliance is also not new, but corresponds to approaches in other instruments, including within the CC. That said, we do see some room for improvement on paragraph 1. We believe that the concept of keeping responsibility with States, parties to armed conflict and individuals should not only be considered across the entire life cycle, but actually.
Be applied to it.
On paragraph 3, one could consider adding "rules of engagement" in the brackets, as they are additional elements that contribute to providing guidance and instructions. On paragraphs 4 and 5, we would have preferred even stronger language on compliance measures For instance, also regarding the imposition of penal sanctions. And with regard to paragraph 4, we would also note that this language is based on amended protocol 2 of the CCW and thus sounds, seems perfectly fine to us as it stands. In addition, the obligation to provide effective penal sanctions and to search and bring to justice those responsible can also be found in the Geneva Conventions and other instruments. Beyond the measures contained in box 5, We would also have welcomed possible statements on cooperation and assistance. I think this could potentially go into the direction of what our colleagues from the US delegation just said concerning the implementation of the provisions contained in the previous boxes or also have additional elements on dispute settlements and clarifications to resolve any problems that may arise. To conclude, another potential element could be the inclusion of some sort of black box into laws which would allow to better understand, reconstruct or investigate the system's actions. Thank you, Mr. Chair.
I thank the distinguished representative of Switzerland for commenting on the text and making proposals. Next on my list is the distinguished representative of Israel, to be followed by the Russian Federation. But first, Israel, you have the floor.
Thank you, Chair. In general lines, we see box 5 in good shape, and we do support the general idea behind it. We just have some comments regarding paragraph 5. This paragraph addresses a highly important subject. At the same time, it exceeds the scope of the work of this group and introduces new terms that have not been subject of thorough discussions. Furthermore, we also note that paragraph 4 is general enough to encompass the practical ideas expressed in paragraph 5. So for these reasons, we think the paragraph should be omitted. If the paragraph is retained in the text, we will suggest replacing the word "must" with the word "should" in order to better reflect its practical nature. Thank you.
I thank the distinguished representative of Israel for commenting and for making concrete proposals. Next on my list is the Russian Federation, to be followed by Peru, Brazil, and Mexico, but first the Russian Federation that we have to.
Thank you very much, distinguished Chair. We also would like to make several comments on box 5. Overall, this section does not lead to any big questions by my delegation. We believe that it's quite close to be agreed. Nevertheless, we need to introduce some polishing up of this paragraph, of this box. Paragraph 2 requires minor amendments to be corrected in a small way, and this is done to make it more clear and more appropriate. And for that reason, we need to include the sentence or the words "employment of" before the word "means," "employment of means," in order to be clear that we are talking about individual accountability for the employment of the relevant means or methods of warfare, as is indeed provided by IHL. On paragraph 3, we believe this paragraph is quite relevant. However, in order to make it more specific, we do not think it is necessary to use very strict language within this paragraph. especially when we refer to certain national measures that could be taken by states in light of the different national practices that are used by different countries. Moreover, for us, it is extremely important to be loyal to the principle of non-interference in the internal affairs of countries. And in this respect, we believe that the measures that are spelled out in subparagraphs a, b, and c, and that could be taken by states at the national level, should not be mandatory. They should be voluntary. And moreover, the list of measures that we see in sub-paragraphs a, b and c, we believe, is non-exhaustive. This represents just some particular examples of measures that could be taken by states at the national level. The range of the different measures that are already being taken at this stage by my country is much broader and is not limited to the measures that we see here. In this respect, we would suggest to replace the word "should" with the word "may". And also after the word "provide", to add the language, inter alia, on a voluntary basis. Moreover, distinguished Chair, we believe that paragraph 5 in this box goes beyond the mandate or the prerogatives of this GGE. And it contradicts the principle of non-interference into the internal affairs of countries. We believe that we cannot impose on states what specific measures should be taken by states at the national level in order to be able to ensure the relevant use of such systems, weapon systems, in line with international humanitarian law. This is a prerogative of each specific country, which types of investigation mechanisms to create or not to create, and how to be able to ensure the responsibility of different individuals that violate the relevant norms and principles and that are under the jurisdiction of that particular State. Moreover, we do have paragraph 4 and the delegation of Israel just referred to that, speaking before us, and we believe that paragraph 4 quite elegantly states that certain measures should be taken by states at the national level, and we believe that paragraph 4 is more than sufficient in order to send the right signal to states in order to take the necessary steps at the national level. This is why we find paragraph 5 superfluous. especially since we have paragraph 4 already. What we have in 5 is covered by paragraph 4, and this is why we suggest the deletion of paragraph 5. Thank you.
I thank the distinguished representative of the Russian Federation for sharing the views of his delegation and for making concrete proposals. Next on my list is Peru, to be followed by Brazil. But let me first pass the floor to the distinguished representative of Peru. You have the floor.
Thank you very much, Mr. Chair. I would like to start by thanking you for this new version of the text of box 5 related to the accountability and responsibility. We believe that the elements that we see here not only make more specific the scope, but also they expand and develop the guiding principles in the meeting of 2019, the guiding principles. And that seek to ensure compliance with IHL and not permit the existence of any lacunae or gaps strengthening not just accountability, but also the responsibility of states. In this respect, I would like to highlight a couple of elements that my delegation finds to be fundamental. Paragraph one, we believe that and we Thank you for keeping the notions of responsibility and accountability. Paragraph 2. Also, we take a positive view, the inclusion of the reference to the human, the chain of human command and control. Paragraph 3, I would like to highlight something that my delegation believes is of particular importance. I'm referring to subparagraph C in paragraph 3, and the importance of training and capacity building of personnel. In paragraph 4, my delegation would like to express its support to the language that we see in this version of the text. This is language that is very similar that was presented by Peru together with some other countries in 2023, working document that referred to a proposal of protocol 6 and in Article 7, it included the different provisions related to compliance and the importance of including such types of reference. And finally, on paragraph 5, my delegation also welcomes the inclusion of this language. And we believe that we can consider the proposal that was made by the delegation of the Red Cross, the inclusion of language related to the development and use. And in conclusion, Chair, my delegation would like to say that we will consider the proposal that was made by the United States related to new paragraph Thank you very much for your attention.
I thank the distinguished representative of Peru for commenting on the tax proposals and also proposals made by other delegations. Next on my list is Brazil, to be followed by Mexico, and then I still have Ireland and Turkey. But let me first pass the floor to the distinguished representative of Brazil. You have the floor, sir.
Thank you, Mr. Chair. On box 5, like we stated regarding the previous box, we believe that the text is very mature and close to a desirable end product. And again, we remain open to changes aimed at enhancing the clarity of the language, but we would not favor major changes to the substance and breadth of this box. In general terms, we believe that there is a need for specific references to individual criminal accountability in the text, as elaborated by the delegation of Bangladesh and others. And we believe that some of the proposals we heard this afternoon go long ways towards addressing this. On paragraph 2, we support the proposal by Portugal and others to insert a reference to traceability. And we also believe that the language suggestion proposed by the ICRC, which is to insert "for humans who plan, decide upon, or execute attacks," is a welcome precision to this text. On paragraph 3, we look favorably towards the proposal by the United States to bring the issue of automation bias to this part of the text. In general terms, we also believe that the proposals made by the US to tweak the language in various paragraphs of this box to insert "their" before "use of laws" is a welcome precision to make it clear that states can only be held responsible for their own actions. We believe such changes could also address the concerns expressed by the US delegation with regard to the paragraph 4, which, like the ICRC, we believe should remain in the text. On the same paragraph, we believe that the general reference to international law should remain as is. On paragraph 4, like Peru before us, we support the proposal by the ICRC to insert a reference to development. And finally, Mr. Chair, we believe that the U.S. proposal on a new paragraph on the voluntary exchange of information is positive, even though there is already an element in this regard in paragraph 1 of box 4. there might be merit in making a more general reference to exchange of information that goes beyond the issue of legal reviews. I thank you, Mr. Chair.
Let me thank the distinguished representative of Brazil for stating the position of his delegation, making proposals. Next on my list is Mexico, to be followed by Ireland, and then I have still Turkey and Colombia. But first, the distinguished representative of Mexico. Madam, you have the floor.
Thank you very much, Chair. If you allow me, taking into account the fact that my comments on box 5 are minimal, I would like to also make some comments on the previous box. At the outset, I would like to thank your team and yourself for preparing this box. We have moved forward significantly on box 4. We'd like to replace the word "should" with "must", and we can also think about the proposal made by Ireland related to obligations and then implementation of practices. On paragraph 1, we welcome the inclusion of the reference to voluntary exchanges between states and independently that subjects of international law involved in particular context of Villigen. Our parties to AP1, the revisions of means of methods of war is crucial to ensure its compatibility with the rules of war. In paragraph 2, we prefer the deletion of the word "significant". We believe that it unnecessarily reduces the scope of the text to place a sufficiently high threshold. Now, I have substantive could be a good replacement with the understanding that the second part of the paragraph highlighting the functions and the effects of autonomous system will be preserved in the text and three paragraph three will believe that the inclusion of reference into human judgment and control across the life cycle of laws is really necessary in this box and we welcome the introduction of a reference to human chain of command and something, a proposal that was made by distinguished delegations. Now, there's all is some duplication. We're a bit flexible in terms of the chapeau four, but perhaps these elements can be reflected in a different way. Now, in paragraph 4, we can support the proposals made by Ireland when we reintroduce the reference to the predictability, explainability, traceability, and others. And also, it's relevant to include a reference to the chain of command and control. In paragraph 5, we agree with Stop Killer Robots when algorithm bias is a better way to describe the problem covered by this paragraph. Also, it'll be necessary to ensure that the text focuses on measures that are to be implemented, will take into account the types of biases and also the situation, the situations that may lead to such bias. And in paragraph six, We support the language presented by the ICRC, and it refers to also cognitive bias, and we believe that it is relevant to highlight in this part of the text that implement measures to prevent and mitigate biases relates to the implementation of the principle of precaution, and this is why we would like to incorporate this principle in this box. Now on box 5. My delegation supports in general this box that establishes a very good balance between the different points that were raised during our discussions. We support the proposal of Finland related to the proposal on voluntary exchanges of states, and we would like to thank the United States for the proposal that was made and the version that will highlight the main points in this part of the text. In paragraph 2, together with Portugal, Brazil and Pakistan, we would like to include a reference to traceability. Now, on the ICR, CPER, we agree with Austria and Brazil. that the proposal will resolve certain concerns raised by delegations, and we support its inclusion in this part of the text. Also, we support the introduction of accountability and responsibility in this part of the text, a proposal made by Portugal. And in paragraph 5, we prefer to keep the word "must," but we can accept "shall." And we accept the proposal of the ICRC to introduce a reference to the development of weapons, something that was supported by the delegation of Brazil and the proposal of Portugal to introduce prosecution. Thank you.
Let me thank the distinguished representative of Mexico for sharing the views of her delegation with regard to box four and five. Let me now pass the floor to Ireland. And then we have still two left, but first the distinguished representative of Ireland, you have the floor.
Thank you, Mr. Chair, and I will be brief. Like others, we also generally support this section as you have drafted, and just a couple of suggestions from our side. We've heard a few delegations talk about the importance of traceability in the section, and we agree as it is crucial for responsibility, investigation, and accountability. So we support the suggestion made by Portugal as it relates to traceability of the chain of human command and control. We would also like to propose the elements of traceability and explainability as they relate to the systems themselves for the reasons that we and other delegations have outlined previously. To that end, we suggest some language that could go as a new subparagraph 3D, which could read, "Measures to ensure that the effects of laws are adequately traceable and explainable in order to permit a reconstruction of events and effective investigation and attribution of responsibility in accordance with applicable international law. This would complement the suggested reinsertion of predictability and reliability in the chapeau of box 4, paragraph 4, if we decide to go for this split approach in reintroducing the four concepts that were in the previous version. Finally, Chair, we join others in supporting the retention of paragraphs 4 and 5 as you have drafted. Thank you.
I thank the distinguished representative of Ireland for sharing the views of his delegation. Next on my list is Turkey, so let me pass the floor to the distinguished representative of Turkey.
Thank you, Chair. I will focus my remarks on paragraphs 2, 4, and 5. On paragraph 2, we propose changing the word "must" with "should" in the beginning of the paragraph. We would like to add a new sentence in the end of the paragraph, which reads, "This is without prejudice to the diversity of domestic legal systems and applicable procedures." Under existing IHL, how states fulfill accountability obligations is left substantially to their decision. To their discretion, apologies. Common Article 1 requires states to ensure respect for IHL, but does not prescribe the specific institutional mechanisms by which they must do so. States retain wide latitude in structuring their chains of command, designing rules of engagement, allocating criminal responsibility among operators, commanders and procurement officials, determining what constitutes meaningful human control in any given operational context. "Must" in the proposed text is followed by a specific operational requirement, namely through the operation of such systems within a responsible chain of human command and control. This goes beyond reaffirming general accountability principles and begins to prescribe specific doctrinal and institutional arrangements. This, in turn, risks encroaching on sovereign prerogatives that IHL has not surrendered. Should preserves the normative direction while respecting states' discretion in implementation. States in this room operate under different legal traditions. Each of us may approach individual accountability differently. The proposed additional sentence in the end of the paragraph helps acknowledge that all of us with differing systems can fulfill accountability obligations through our own legal systems and procedures. Now, moving to paragraph four and five, I'll be short because we share concerns raised by several delegations before me. Like them, we call for deletion or we request replacing the words in those two paragraphs, the words "must" in those two paragraphs with "should" and inserting "applicable" or "as applicable" in appropriate parts of each paragraph. Thank you.
Let me thank the distinguished representative of Turkey for sharing the views of her delegation. Next on my list is Colombia. You have the floor.
Thank you, Chair. Briefly, we're taking the floor to defend paragraph 4 of this box 5 as you presented it. This paragraph makes states prevent and suppress violations of international, including IHL. And for us, this paragraph represents a fundamental part of respect, not just IHL, but also in general of international law and the regulation of these weapons systems. The caveats that we listened that were referred to by other delegations should not be included in this paragraph. And also, we'd like to say that it is important to include both reference meaning reference to IHL and to international law, since the obligations of prevention, investigation, and suppression are not simply limited to armed conflict and are not limited exclusively to IHL. This is something that can happen during the stages of development, deployment, And we have to think about the threshold of the application of IHL and also the proposal made by the United States related to voluntary exchange of information, something that we could support. Thank you.
I thank the distinguished representative of Colombia. I have now also Cuba, so two more. We got a few bonus minutes from the interpreters, but it would be very helpful if we will be able to indeed round up our discussions with regard to box number five. So I would now like to pass the floor to Austria and then to Cuba. Thank you.
I will be very quick, just a reaction to the proposals that we should change a must to a should when it comes to para 2. I just want to refer again to the 2019 report of the CCW, of the GGE on laws, para 17C, where we used this kind of formulation and back then we used must. So I don't think that we should go backwards and go from musts that we have already agreed to shoulds. I think that would be the wrong direction. Also on para 4, should would not make a lot of sense because I don't think that we would all agree that states should prevent and suppress violations of international law. I think that it's very clear that they have obligations to suppress and prevent violations of international law. Thank you.
I thank you for your brevity. And last on my list is the distinguished representative of, well, then we're not going to make it. Okay, Cuba and then Canada.
We'd like to express our support what was said by countries that are in favor of keeping paragraphs four and five as presented by you, Chair. Thank you.
Next on my list is Canada.
Thank you, Chair. Like others, my delegation is largely supportive of box five in its current form. We have listened carefully to the proposals made by other delegations on this box and just wanted to share a few additional remarks. We believe that this box focuses on the measures of accountability which are necessary to ensure the lawful use of laws. Turning to paragraph 3, regarding the suggestion to move the autonomous bias element from box 4, paragraph 6, we could look favorably on this as it is a measure that relates to human responsibility and accountability. In this regard, we believe that automation bias should still remain in box 4 as a measure required to ensure the safe and responsible development and use of laws in compliance with IHL, given it relates to the interface between humans and the autonomous system. In Paris 4, we think that the term appropriate is ambiguous and subjective, and we would recommend using the term necessary as a suitable alternative. Furthermore, the phrasing persons or territory under jurisdiction is language commonly associated with international human rights law. The obligation here is to respect and ensure respect of IHL and to prevent violations. That obligation attaches to states and commanders, not machines. And finally, Mr. Chair, in paragraph 5, we agree with Sweden, Portugal, and others that should instead of the current must is the term that we would like to see, as we're not referring to a legal obligation. And just to say that regarding the calls for a new paragraph on informal exchange, like other delegations, we could also look favorably on this suggestion, as that could be an important additional measure to promote the accountable and lawful use of laws. Thank you.
I thank the distinguished representative of Canada. And then the last on the list, and I think we are going to make it then, is the ROK. So you have the floor.
Thank you, Chair. I don't want to keep everyone from going home. I am the one who wants to go home. But just to comment two things. First, traceability. I think there was in the previous version, there are many delegations objecting the traceable command and control, because if the argument is that the system is not traceable, there's no point putting traceable in front of the human command and control, because the chain of command and control is always traceable. But we can look positively about adding new paragraph on the traceability and explainability as one of other possible measures in other paragraphs. And Second point on development. I think our common understanding is that we have a separate paragraph on the development design or the life cycle in other paragraphs. We don't want to go back to put development in any other, every other paragraphs, either in this box or other boxes. Thank you.
Let me thank the distinguished representative of Korea for reacting to some of the proposals made. I also thank the interpreters for giving us just a few more minutes because it's helpful that we can indeed close the discussion now on box 5. I'm not going to summarize the proposals. I think the main issues are indeed, I think, with the boxes 4 and 5 because we have proposals for the lesion there. And the rest, I think, is I would not say cosmetic, but is linguistic and is preferential. With regard to our work tomorrow, we will do some further work again this evening with regard to previous boxes. and then have a selection of that discussed tomorrow during the day. And tomorrow evening, I would like to dedicate to the issue of human involvement and the famous abbreviation we have for that in the text. I think that is what I can say you now. I would like to thank you all for your constructive engagement today. I wish you a pleasant evening, and I see you tomorrow morning at 10 o'clock in the room. The meeting stands adjourned.