3rd Meeting - 1st Session Group of Governmental Experts on Lethal Autonomous Weapons Systems 2025

Good morning, Excellencies, distinguished delegates. Day two, box two, second-half of the discussion. We still have at this moment one, two, five, we see Sweden, six, Sweden. I contracting parties asking for the floor to continue our discussion on box two. I have first Australia to be followed by Mexico, Finland, the Republic of Korea, Cuba, Sweden. But let me first pass the floor to Australia. You have the floor.

Thank you, Chair.

We are pleased that the majority of this section appears to enjoy a level of support in this room. Chair, Australia supports the inclusion of context-appropriate human control and judgment in paragraph five as a pragmatic compromise. However, we are aware that other states have expressed strong views concerning meaningful human control as an alternative.

Australia's concerns with terminology like meaningful human control is well known.

We consider this phrase vague and subject to differing interpretations. It is therefore not overly helpful in clarifying how IHL applies to laws.

Or giving concrete guidance to states on how to ensure compliance with IHL. We would propose that this paragraph is better suited in section four.

Section two concerns the application of IHL to laws, and context-appropriate human control.

And judgment is not an existing IHL norm.

Concerning the use of the term employment.

We think it would be amenable to many in this room to replace employment with use in this context.

We are open to the use of the terms development and use.

But we think that the insertion of this phrase needs to be assessed on a case-by-case basis.

I thank you.

Let me thank the distinguished representative of Australia for her intervention and suggestions. Next on my list is Mexico, to be followed by Finland. Let me now pass the floor to the distinguished representative of Mexico.

Thank you very much, Chair. Good morning, colleagues. My delegation at the outset would like to join those that believe that in general the text of this section two is quite adequate. And we also do support the amendments that were suggested by Pakistan, Switzerland, ICRC and other colleagues on different aspects. On paragraph 4, we believe that it is important that we adjust ourselves to the Martens clause and eliminate in this respect the reference involving laws not covered by the CCW and its annex protocols in the beginning of the paragraph. On paragraph 5, my delegation supports the use of the terms development in use in line with the provisions of IHL applying not only to the use of force as we have explained earlier. Nevertheless, we believe that in order to complement that idea and be fully aware of our obligations under international law, it will be appropriate to include the paragraph three bis. That will include the first part of paragraph one of section four related to the obligations under article 36 of the Additional Protocol for many delegations. This obligation is part of our legal commitments that are applicable in this context, and it'll be better to include that in section two. This means that the sentence compliance with obligations under international law already recognize that additional protocol one is universal. We also support delegations that were in favor of use of the terms control and human judgment in appropriate context. I believe that the explanations that were given by Brazil and Switzerland is something that we can support. Finally, my delegation can support the proposal made by Colombia to have a reference to international law in general, but as an alternative, my delegation can suggest another paragraph 5 bis to see what is the position of delegations that states the previous paragraphs are without prejudice to obligations deriving from other international legal regimes, for example, international human rights alone and international the law on international responsibility. Thank you.

I would like to thank the distinguished representative of Mexico for his intervention and suggestions and comments on other proposals. Next on my list is Finland, to be followed by the Republic of Korea. But let me first pass the floor to the distinguished representative of Finland. You have the floor.

Firstly, I'd like to shortly state our overall support for the second box, especially the second paragraph on IHL appliance. which we find to be a highly important element of the rolling text. When it comes to para four, we see merit in including the Martens clause as the ethical considerations and IHL are relevant in assessing autonomous weapons systems. The Martens clause is customary IHL and has also been adopted in many international treaties. But in order to clarify the text, we agree with delegations proposing that the first part of the sentence referring to CCW and other international agreements could be deleted. About the fifth paragraph, we have preferred to use context appropriate human involvement, similarly to what was mentioned by the UK yesterday. However, we can agree with the current formulation of context-appropriate human control and judgment, which we understand to be a compromise drawn from our previous discussions regarding the essence of necessary level of human control. For us, context-appropriate is a necessary caveat when using the words human control instead of referring to human judgment or human involvement. Thank you.

Let me thank the distinguished representative from Finland for her comments. Next on my list is the Republic of Korea, to be followed by Cuba and Sweden. But let me first pass the floor to the Republic of Korea. You have the floor.

Thank you, Mr. Chair. Regarding box 2, my delegation has also three points. First, my delegation aligns with the majority of the views expressed in this room that box 2 is in good shape. We understand that paragraphs one through four reaffirms our commitment to existing principles and norms, which are largely acceptable to my delegation. Secondly, it seems, however, we need more discussion on paragraph five, particularly the term context appropriate human control and judgment, which we consider the core problem here, or we see it as the lethality of box two. As demonstrated in yesterday's discussion and today, there is no consensus on this long and complex term. To better reflect our past discussions and common understanding, my delegation agrees with the U.S. delegation that the phrase should be control and human judgment, or we suggest human judgment and control as an alternative. Lastly, we support the views expressed by the U.S., India, Australia, and others that the blanket use of the term development and use regardless regardless of the context throughout the text is not wise. A more tailored approach is necessary. Thank you.

I would like to thank the distinguished representative of the Republic of Korea for stating the positions. Next on my list is Cuba, to be followed by Sweden and Portugal. But let me first pass the floor to Cuba. You have the floor.

Thank you, Chair. In a general manner, we agree with the text, but would like to make several comments. Paragraph one, after war, we suggest we eliminate "is not unlimited," but replace it with a phrase that will give consistency to the box, that it can be governed by international law, including international humanitarian law. On paragraph 3, we can modify it to say that international law, to add it, and we can say it applies fully to the entire life cycle of laws. Now, paragraph 4, as a result, will not be necessary because it will be summarized then in the idea contained in paragraph 3. In any case, I'll be sending my proposals to the Secretariat to facilitate your job. And we are at a stage in our discussions, perhaps it will be not necessary to have in paragraph two the reference to methods of war, those of the past, those of the present, those of the future, because what we're seeing today is development of technology and evolution of armed systems. That is already happening. Thank you.

I would like to thank the distinguished representative of Cuba for stating her views and positions. Next on my list is Sweden, to be followed by Portugal and the campaign to stop killer robots. But first, let me pass the floor to the distinguished representative of Sweden. You have the floor.

Mr. President, Sweden supports the paragraphs in box 2, but we want to respond to some of the statements put forward by delegations yesterday and this morning. We echo the encouragement to be consistent in terminology, such as employment or use, but we note that there may be places later in the rolling text where we only want to address use versus employment and not development. In relation to the discussion on para four, Sweden would like to point out that different versions of the so-called Martens Clause appear in several IHL treaties, for example, in the Geneva Conventions of '49, to which all states in this group are parties to. Along that line, we believe that a mere reference to principles of international law derived from custom insufficiently reflects modern IHL, which is built up by both treaties and custom. and contains rules in addition to principles that could be applicable to systems potentially falling outside the scope of these elements. So we would therefore suggest a slight reformulation of the para just to make it more stringent. The para would sound as follows, and we will of course send our suggestion into the Secretariat afterwards, but I will read it. In cases not covered by the present instrument, the civilian population and the combatants at all times remain under the protection and authority of the rules and principles of international law, in particular IHL, derived from applicable international agreements, established custom, the principles of humanity and the dictates of public conscience. The notion of applicability could serve as a reminder that States have different treaty commitments and that treaties have different applicability clauses. Moving on to para 5, we support this para, but we could also consider refined formulations that capture its essence. And I want to point out two key aspects for us in this regard. For Sweden, it is key that laws can be used in line with IHL. Further, it is a key objective for Sweden to preserve human control over the use of force. Measures to ensure such control could be considered throughout the entire life cycle of a weapon system. Specific measures must, however, be context dependent and assessed on a case by case basis. In light of this and different views put forward over the years, we believe that the term context appropriate human control and judgment captures both the need for meaningful human control and the need for adaptability to the context and arena of operations in a formulation that could be suitable for this type of elements. Thank you, Chair.

I would like to thank the distinguished representative of Sweden for sharing his views and making comments. Next on my list is Portugal, to be followed by Campaign to Stop Killer Robots, France, Italy, Sierra Leone, and then I have also Pakistan, El Salvador, and Germany. But let me first pass the floor to the distinguished representative of Portugal. You have the floor.

Thank you, Mr. Chair. In general, we support the rolling text, which is essentially based on agreed language. We support the use of the concept of human control in para five and throughout the rest of the text. Such concept implies accountability and acts as a means of overcoming possible system failures. We do not oppose to the addition of the term judgment, even if we do not consider it strictly necessary. Also, in para five, We consider that although not explicit, international law applicable to laws is not limited to IHL, but includes international criminal law and international human rights law too. And that would be all for now. Thank you, Mr. Chair.

Let me thank the distinguished representative of Portugal for his intervention. Next on my list is Campaign to Stop Killer Robots. You have the floor.

Stop Killer Robots wishes to thank the chair for your work in organising and chairing this meeting and for your development of the chair's rolling text. Stop Killer Robots welcomes this approach and sees the text as useful and concrete progress towards a legally binding instrument. With regards to box two, paragraph five, and the suggestions by some states that we might modify or remove the reference to context-appropriate human control and judgment, we would like to urge states to keep this text, which clearly links compliance with international law with human control over weapon systems. While we believe that meaningful human control offers the clearest expression of what is called for, we recognize the views of states that judgment is also important in the context of responsibility and accountability. We also recognize that appropriate or context appropriate may seek to express the same substantive content as meaningful, though these expressions may also carry more specific meanings in the context of state policies. While we encourage States to find consensus on the best way to express this idea, we want to reiterate the view of the Stop Killer Robots campaign that the most essential criteria for this paragraph is to ensure the centrality of the human element. While existing IHL does not explicitly refer to the need for human control, it is precisely because of the autonomous nature of these technologies that human control over them needs to be explicitly stated and required. We also understand the need to clearly express the nature of the control and/or judgment and believe that the text of boxes three, four and five aim to articulate and flesh out what constitutes the key, the kind of human control needed. But those texts become meaningless and this text loses the requirement for human control over weapon systems. Thank you, Chair.

Let me thank you for your intervention and your comments and proposals. Next on my list is France, to be followed by Italy and Slovenia. Let me first pass the floor to the distinguished representative of France. Madam, you have the floor.

Thank you, Chairperson. Overall, we agree with box 2. This could be It is a useful element which we could accept as it stands, but we would like to come back to certain comments that have been made by delegations yesterday and this morning. In paragraph 4, like Finland, we could support removal of the first part of the sentence referring to the preamble in cases not covered by this convention.

And by other international agreements.

And the rest of the sentence would remain. In paragraph 5 we also felt that the concept of human control and judgement included the concept of judgement but we have heard a number of delegations on this in a spirit of compromise we could accept the expression as it stands context appropriate human control and judgement we also accept the addition of the notion of development in this section, which makes it clear that human control should be present throughout the life cycle. Like Australia and Sweden, we will view on a case-by-case basis the concept of the idea of including development in other parts of the text. Thank you.

Let me thank the distinguished representative for France for sharing her positions and views. Next on my list is Italy, to be followed by Sierra Leone and Pakistan. But let me first pass the floor to Italy. Madam, you have the floor.

Sorry, Chair. The application of international humanitarian law to laws is at the core of our discussions and one of the most relevant aspects that would be addressed by the future instrument. Italy is supportive of the current version of this section. In our view, existing international law, in particular, the general rules of IHL governing the choice of means and methods of warfare, and specific treaties and customary rules prohibiting and restricting certain weapons, fully apply to new weaponry. We consider therefore essential that we address the application of IHL through a technology neutral approach resistant to the constant and rapid evolution in the technological field. In paragraph 2, we think that the exclusive reference to military technology used risks to exclude, in the future, technologies of dual use nature. It is a point also touched upon yesterday by the distinguished delegation of Japan. In line with the paragraph 23C of the 2019 GG report, which recognizes the dual use nature of the underlining technologies in the area of lows, we rather prefer independent of the technology used. We are equally content with the language suggested by the distinguished US delegation independent of the specific technology. In addition, in paragraph five, we would like to add, as already flagged by the distinguished delegations of Pakistan, Brazil, the ICRC and France just now. Development before employment, in line with Article 36 of the first additional protocol to the Geneva Convention, according to which each state is required to determine whether the employment of a new weapon means and methods of warfare that studies, develops, acquires, or adopts, would in some or all circumstances be prohibited by international law. We in fact believe of paramount importance to preserve and retain human control at all times and across the entire life cycle of the weapon, including the political direction of the pre-development phase and the design of the weapon. Finally, Chair, looking at this rolling text as a wall, We saw references in some sections to use, in others to employ employment, and in others more to deployment and use, development, deployment and operation. We are flexible, but to ensure consistency across the text, we recommend using a single terminology, preferably development and use in a consistent manner. We consider these two terms, in fact, to encompass also deployment, employment, and operational laws. And I thank you, Chair.

Let me, Frank, to distinguish. delegate from Italy for sharing her views and opinions with regard to the text. Next on my list is Sierra Leone, to be followed by Pakistan, El Salvador and Germany. Let me first pass the floor to oh, and I have already Denmark and Norway. Let me now pass the floor to the distinguished representative of Sierra Leone. You have the floor.

Thank you, Mr. Chappasin, for giving the floor to Sierra Leone.

My delegation fully supports the text in this box.

Sierra Leone strongly supports the affirmation that human control and judgment is essential for compliance with international law, especially IHL.

Principles of distinction, proportionality and precaution in attack.

If context appropriate proves to lack consensus, we recommend deferring to our previous recommended language as stated by previous speakers from the G16 group, meaning.

We believe this is a crucial part of the ruling text, therefore it must be retained.

As has been highlighted by previous delegations on this issue, and in order to ensure clarity and consistency, we recommend using the terms development and use consistently, therefore, the document instead of varying terms like employ, deployment or operation. While we therefore, while we, sorry, while we welcome the restatement of IHL principles, Sierra Leone believes we must move beyond merely restating existing rules to elaborate specific prohibitions and restrictions addressing the unique challenges posed by autonomous weapons system, as illustrated in the upcoming boxes. Finally, we further support the recognition of the Martens Clause in paragraph four, ensuring that civilians and combatants remain protected by customary law, humanity principles, and dictates of public conscience. in cases not explicitly covered by existing agreements. Thank you very much.

Let me thank the distinguished representative of Sierra Leone for sharing the views of his delegation and making proposals. Next on my list is Pakistan, to be followed by El Salvador and Germany, but let me pass the floor to the distinguished representative of Pakistan first. You have the floor.

Thank you, Mr. Chair, for giving us the floor for the second time. We appreciate the interactive nature of the discussion and hence our second intervention. And we would like to respond to some of the comments that were made yesterday following our intervention on box two on two issues, specifically the development and use and human control. First on the issue of development, Mr. Chair, We have listened carefully to the views of certain delegations who have argued that IHL is primarily concerned with the use of weapons and that a violation occurs when a weapon is used and not as much as it is developed. So they suggest that our regulatory efforts should be focused solely on the use and effects rather than development. What these delegations refer to is the body of targeting law within IHL, the set of rules governing how weapons are used in hostilities. This is a view, but we do have another perspective and we differ from that. And the perspective, we believe, what these delegations point out overlook a crucial and well-established legal principle that ensuring IHL compliance must begin at the development stage, and right now many delegations have pointed out, delegation of Italy also just pointed out that Article 36 of the Additional Protocol I of the Geneva Conventions do affirm the necessity of ensuring compliance with IHL from the Inception, inception stage of a weapon cycle, the very purpose of legal review under article, article 36 is to prevent the development of weapon that cannot be un, uh, cannot be lawfully used in, in an armed conflict. Others may argue that Article 36 is not a customary obligation, but the reality is that many of these States, including those that are not party to Additional Protocol I, already conduct legal reviews of new weapons as a matter of national policy. We respect that national policy and the practice is worth emulating. And this does reflect a broader recognition that a weapon compliance with IHL is determined long before it is deployed or for that matter employed on the battlefield. Mr. Chair, beyond targeting law, IHL also includes weapon law, a body of rules that establishes prohibitions and restrictions on specific weapons, means, and methods of warfare. This pertains not only to how weapons are used, but how the nature and design of the weapon system, their inherent behaviour and how they affect people or objects against which they are directed. And our fundamental question does remain, how do we address weapon systems that by their very design or development parameters cannot be used in compliance with IHL. I think this is a question that would beg an explanation if we were to omit development. And we provided specific example. For example, if bias is not eliminated at the design stage, the system may inherently discriminate against protected person in armed conflict, and if predictability, explainability and reliability are not ensured, at the development phase, the weapon may operate unpredictably, increasing the risk of violations of distinction and proportionality. The laws of war are not merely to react to violation after they occur, they exist to prevent unlawful harm. So the element of prevention will also be left wanting if development stage is not addressed. A weapon system that cannot distinguish between combatants and civilians cannot apply proportionality attack and cannot operate with necessary human oversight, and that cannot suddenly be made lawful by simply regulating its use. Secondly, Mr. Chair, on human control and judgment, we heard the argument that the phrase of human control is too narrow and that IHL does not explicitly require human control, that Previous GGE reports have not established human control as a consensus term. We also heard the view that human judgment alone may be sufficient to ensure oversight and that we should simply reproduce the agreed language from six years ago, as contained in the 2019 GGE. Our response to that is that we have already established in 2023 GGE report that control is essential for upholding compliance. This is not a theoretical debate, it is a reaffirmation of a fundamental legal and operational requirement. The question before us is not whether control is necessary, but rather who or what exercise this control. Can compliance with IHL be ensured without direct human control, solely through an autonomous system? Or does IHL require human control over critical functions to ensure lawful decision making involved? The essence of IHL is predicated on human agency. It is a legal framework built on the assumption that humans, not machines, make decisions about life and death in an armed conflict. Weapon systems with autonomy and machine learning, no matter how advanced, cannot always replicate the nuanced decision-making required in the complexities of war. IHL's core principle, distinction, proportionality, and precaution, demand context-sensitive reasoning that AI-driven system cannot guarantee fully. Human control is not just about compliance, it's also about preserving the very structure open upon which the IHL is built. If human control is absent, this raised an unresolved and deeply troubling question, who is responsible for the attack if no human meaningfully controlled it? IHL is quite clear, Mr. Chair, legal and moral responsibility for an attack must be assigned to a human operator or a commander. But if an autonomous weapon system engages targets on its own, where does the accountability lie? And this has been the debate going on. Explicit terms, human control, another, it is true that explicit term of human control is not found in existing IHL treaties. But here I think we need to really apply our mind and focus on at which point of international legislation so as to speak we are on. The question has never arisen before, rightly. Because the weapon system that thus far have been used have relied on human decision making at critical moment. The fact that IHL does not explicitly state human control is not because it is unnecessary, because it has always been assumed as such. The emergence of laws forces us to ask a question that was never previously considered. As such, this is an evolutionary aspect, an evolving form of international humanitarian law. So in future, if, for example, AI or emerging technologies will impact other aspects of IHL positively, for example, in the cases of protection, if AI application were to aid to the protection and the responsibility to protect, then we will have to refer to those aspects positively for a human context. So the broader question, do we allow machines to replace humans in making life and death decisions? IHL has evolved before and it must evolve now to ensure that its core tenants remain intact. Finally, I am sorry if I have been too long, but I will make one point on human judgement. It has been said that human judgement alone is broad enough to ensure oversight, but judgement without control is meaningless. A commander may make judgement, but if they have no ability to intervene or override an autonomous system's action, then the judgement has no legal or operational effect. Judgement must be accompanied by control. Otherwise, we are just creating an oversight, an illusion of oversight, and that's where I stop. Thank you very much.

I would like to thank the distinguished representative of Pakistan for sharing his views and in a way also already linking the discussion we have now on box two with text that is in box five. As stated before, it's all interrelated. We go box by box, but it's always good to to be aware of those relations in the text. Next on my list is El Salvador, followed by Germany, but let me first pass the floor to the distinguished representative of El Salvador. You have the floor.

Thank you, Chairperson, and good morning, colleagues. On box two, may delegation generally supports the proposal on the screen and once more we appreciate the efforts made by the team. We support the statements by various delegations yesterday including the G77 and we would like to express our support in particular to the issues mentioned during this debate. Firstly, seeking to be consistent with the paragraphs, we support replacing employment or deployment by development and use. It is important for us to take into account the entire life cycle of this weapons system. On paragraph 5, my delegation prefers the term meaningful human control and supports the explanation given by Pakistan on human control. But we could also consider what has been stated by Brazil and Mexico. In conclusion, my delegation supports Colombia's proposal regarding international law and the five bis mentioned by Mexico. Thank you.

Let me thank the distinguished representative of El Salvador for sharing positions and his views on the proposals of other delegations. Next on my list is Germany, to be followed by Denmark and Norway. But let me first pass the floor to the distinguished representative of Germany. Madam, you have the floor.

Thank you, Mr. Chair. Germany fully supports the remarks on the role of international humanitarian law contained in part two of the rolling text. Given that the use of new technologies in the area of laws may lead to enormous changes in warfare, we consider it crucial to reaffirm that IHL governs the use of all weapons, including those of the future, and that its applicability is independent from whatever technology may be used. Given the core importance of IHL compliance, we see a substantial added value in including a reaffirmation of the applicability of IHL in the area of laws in possible elements of an instrument. Our previous discussions, including the discussions that we had yesterday and this morning, have demonstrated that one of the most difficult questions remains how to describe the role of the human with regard to laws. After so many discussions on this question, we are convinced that we have the same goal to ensure that weapons systems will be employed in compliance with international law. Accountability and responsibility need to be preserved. The human needs to play an appropriate and significant role that goes beyond meaningless involvement. It is our impression that our discussions are on the right track. Human control is at the core of Germany's national position on laws. We refer to the necessity of maintaining a framework of human control and we welcome the Paris Declaration on maintaining human control in AI enabled weapon systems that was endorsed on 11 February and underlines the meaning of the concept of human control. The wording contained in the rolling text does not fully reflect our national position. Nevertheless, in the spirit of pragmatism, Germany supports the suggestion made by you, Mr. President, namely context appropriate human control and judgment'. In our view, this wording sufficiently reflects the core aspects that need to be maintained, in our view, with regard to development and use of any weapon system. Thank you.

Let me thank the distinguished representative of Germany for sharing views and positions of her state. Next on my list is Denmark to be followed by Norway, Peru and Canada, but let me first pass the floor to the distinguished representative of Denmark. You have the floor.

Thank you, Mr. Chair. Denmark is generally supportive of Box 2. We have some minor suggestions for language tweaks that we have sent to you and allow me to explain them shortly here. In paragraph two, we would suggest to delete the words,

those of the past, those of the present, and those of the future, as the applicability of IHL

to past, present, and future weapons means of and methods of warfare is an embedded part of IHL. However, we are flexible on this and understand the intent to underline the governance of future technology. In paragraph three, we also support a reformulation of the sentence so it reads, IHL applies fully to the development and use of laws. In paragraph four, we find merit in finding a formulation that captures the Martens Clause and look forward to further studying the proposal from Sweden. Lastly, I would like to highlight that we find it important to capture the concept of human control in this paragraph. and we can support the formulation context appropriate human control and judgment in paragraph five. Thank you.

I thank the distinguished representative of Denmark for sharing the views of his delegation. Next on my list is Norway, Peru, Canada, but let me first pass the floor to the distinguished representative of Norway. You have the floor.

Thank you, Chair. We too are generally supportive of box two. In this intervention, I would like to focus on paragraph 5 and the concept of human control, which in our view is the very core of the law's discussion, because it is precisely the capacity of autonomy to change and minimise the role of the human operator that presents us with these legal and ethical challenges that we grapple with. We quite appreciate both the inclusion of paragraph 5 and the way that it is formulated here. Unlike the other instances where human control is mentioned in this text, this paragraph draws the line between human control and compliance with IHL and shows the intrinsic connection between the two that in order to comply with IHL, you have to have some kind of control over your weapon. By 2019, this GG had concluded that human judgment is necessary, essential, as it were, to ensure that laws be compliant with IHL, and further that IHL, including the principles of precautions in attack, proportionality and distinction, must be applied by humans in a responsible chain of command and control, that's guiding principles D and E. In other words, humans must make certain judgments in relation to the context in which the weapon is used and subsequently apply these, which in turn requires that you then have some kind of meaningful control over your weapon, otherwise you won't be able to give effect to the outcome of those judgments. That is also further underscored by the previous acknowledgements that this GGE arrived at regarding the need for human accountability and responsibility. In 2023, of course, the GGE went a step further and concluded that control with regard to laws is needed to uphold compliance with international law in general and IHL in particular. We see paragraph 5 and the various references to context-appropriate human control throughout the document as the next small but logical step on this journey. From a national standpoint, we would prefer the formulation meaningful human control', which to us has become quite clear over the years, at least the core meaning of it, whilst of course it still remains a bit fuzzy around the edges, but we accept that that concept has a lot of baggage, and the context appropriate human control and judgement' that is introduced here is, in our opinion, a fair alternative. It underscores that there has to be a certain quality and scope to the control first, that the user must be able to influence the system in a non trivial way, and second, that those who are in a position to influence the system, whether the operator or the chain of human command and control more broadly, must exercise some level of judgment, be that legal or otherwise, in relation to the context of use, that they are not in a way, just some automatons pushing buttons when prompted. Um, on the flip side, there is nothing in that concept that suggests that humans could not be assisted by software in the exercise of this control. Um, it was suggested yesterday that human judgment is part and parcel of human control and hence that mentioning judgment here is redundant. We agree, yet in the absence of a qualifying adjective like meaningful', we think it makes sense to underscore that proper control necessarily is linked to human judgement, and in order to highlight this relation, if that is really necessary, we could perhaps have written human control, including human judgement', but we don't think that is necessary and it would make for a rather clumsy sentence, so we would rather just stick to the current formulation. Finally, it was proposed yesterday that we avoid the concept of context appropriate or meaningful control entirely and focus rather on identifying the measures to be included in box 3, the rationale being, if we understood correctly, that we are in a sense putting the cart before the horse, that we should not try to establish a concept which is not legally recognised and not entirely clear, from which we then start to derive various measures. that instead we should first work on identifying the measures and then, at some later stage, try to find the label that captures the essence of those prohibitions and regulations. That's a fair point, but we think that there is an argument for having both. There is an interplay between the two. The concept of human control, in whatever way we describe it, gives direction to our work. It highlights that very special and, in a sense, uncomfortable aspect of autonomy which distinguishes laws from classic weapon systems and which is the very reason that we discuss the need for prohibitions, regulations and other measures, and it also has an expressive function. Conversely, the closer meaning and content of human control will ultimately be determined by the concrete measures that we then proceed to identify. A word on the issue of employment versus development and use as well. We agree that weapons must be developed in such a way that meaningful human control is possible. Nevertheless, the way that paragraph 5 is structured now, it would not make so much sense, we think, to replace employment with development and use, because if we did change it that way, we would be saying that context appropriate human control over the use and effects of a weapon would be necessary to ensure that it is developed in an IHL compliant way, but it's the other way around in order to ensure IHL compliance, laws need to be developed in such a way as to ensure human control over the use and effects. We would be open to including that somehow, and we think that Italy and Pakistan argued the point very well. Our point is simply that we cannot do it in paragraph 5 through a simple search and replace. One final thing regarding the Martens clause, we found the Swedish proposal interesting. We will have to chew a bit more on it, but we think that it has merit. Thank you.

Let me thank the distinguished representative of Norway for sharing his views and elaborating on the issue of the relation between human control and judgment, as a follow-on also to the earlier remarks made by the distinguished delegate from Pakistan. And yesterday we had also some remarks from the UK already how that nationally is implemented. I think this is all very helpful to better understand how nations view these terms and what they would mean in practice, because we are not just developing text, we want to do something with it. Next on my list is Peru, to be followed by Canada and Bulgaria, but let me first pass the floor to the distinguished representative of Peru. Sir, you have the floor.

Thank you, Chair. On the elements from box two, that includes reaffirmation of international humanitarian law and under fundamental principles contained in IHL, my delegation welcomes. And we see that some of the conclusions from 2023 are reflected in this box. In paragraph three, as was mentioned by some delegations, my delegation believes that we have to include a reference to international law in general, and also a reference that all of this is without prejudice to the applicability of other branches of international law, for example, international human rights law, international criminal law, and international responsibility of states. Also, in order to be consistent throughout the text, we believe, just like some other delegations, that would prefer to use the same terminology throughout the text. In other words, development and use. And when we refer to the word development, in light of Article 36 of Additional Protocol I of the Geneva Conventions, indeed it talks about that during the study development acquisition or adoption of new weapon systems. The high contracting party has to determine whether its use will be banned by the protocol or any other norm of international law. This was very clearly explained by distinguished delegation of Pakistan just a few moments ago. And finally, along the same lines, in light of the reference to meaningful human control, I would like to say that Peru is one of the countries of the G16 and the proposal of Protocol 6 additional. There is a reference to meaningful human control in that proposal. In any case, we believe that it's important to keep language at this point of our discussions as we exchange our views. We need to preserve in this box a reference to human control, as already was mentioned this morning. Any judgment without control doesn't make any sense, is meaningless. Thank you.

Let me thank the distinguished representative of Peru for sharing the views of his delegation and also commenting on previous interventions. Next on my list is Canada, to be followed by Bulgaria and Spain. Let me now pass the floor to the distinguished representative of Canada. You have the floor.

Thank you, Chair, and good morning, colleagues. On box two, I can be brief. My delegation agrees with the paragraphs as they are presented, and we note that paragraph five has provoked the most discussion and debate in the room. We've listened carefully to the interventions made on the concept of context appropriate human control and judgment. My delegation is comfortable with this formulation since, at its essence, what we're saying is that humans cannot transfer responsibility and accountability to a machine. We could also propose context-appropriate human accountability and judgment, of which might be able to address the concerns regarding the use of the term control. If the term control is to be retained, we would also like to propose accountable human involvement with context, appropriate control and judgment. Thank you.

Let me thank the distinguished representative of Canada for sharing the positions of his delegation and coming with proposals. Next on my list is Bulgaria, to be followed by Spain and Austria. So you have the floor.

Thank you, Chair. Mr Chairman, Bulgaria supports box two as a whole. We have some brief remarks. In para four, we see value in adding the Martens clause. However, similar to our other delegations, we would prefer the deletion of the first part of the sentence. Another option for us would be the Swedish proposal for this para. On Para 5, we believe that the core and the essence of the group's discussions should be the human-machine interaction and the concept of human control and its applicability in compliance with IHL. We support the inclusion of context appropriate as a useful and pragmatic compromise. For us, it is important to note that human control can be described as context dependent. And the extent and type of human control to be exerted over an autonomous system to guarantee compliance with IHL depends on the complexity of the operational environment, the intricate characteristics of a weapon system itself, and its capabilities and capacities, and also on its intended use and task to be performed. Thank you, Chair.

Let me thank the distinguished representative of Bulgaria for his comments. Next on my list is Spain to be followed by Austria and the Philippines, but let me first pass the floor to the distinguished representative of Spain. You have the floor.

Thank you very much, Chair. I'd like to thank you at the outset and thank your team for the work you carried out and the way you're guiding our discussions. We're very pleased that we have good discussions here and also we're very pleased with the broad participation of delegations all of this with a view of fully implementing our mandate. Spain expressed its support to the paragraphs included in box two since it reflects and translates perfectly what was discussed previously in our meetings and also it's fully in line with the guiding principles that were previously agreed on paragraphs one, two, three. We agree with those paragraphs and we would like to keep the current language because it's fully in line with the norms of IHL and the guiding principles of 2019 on paragraph four. We believe that it'll be good to have an explicit reference of the Martens Clause to humanity and reflect what is contained in the preambular part of our convention and also in the Geneva Conventions. Now in paragraph five, Spain has always been in favor of human control throughout the entire life cycle of these weapons. We also think that it is adequate to have a reference to context appropriate because the degree of control will depend on the following factors, the type of system where it's being employed and others also for us is crucial to ensure that reliable judgments can be made based on the information available at every point. The principles of distinction, proportionality, and precaution in attack should be applied throughout the entire life cycle of these weapon systems. Thank you.

Let me thank the distinguished representative of Spain for sharing the views of her delegation and also reacting to the proposals made during previous interventions. Next on my list is Austria, to be followed by the Philippines, Costa Rica, Japan and Greece. But first, the distinguished representative of Austria, so you have the floor.

to apologize to take the floor a second time, but it's a very interesting debate and we took up many important elements that have been brought up, especially by the excellent interventions by Pakistan and Norway. I think what we have to keep in mind when we talk about box two and this section is that this is supposed to be a possible preamble or part and it's laying out some general principles. Many of them are, as, as Norway has explained, developing from, from elements that we already have in, in previous text, including the 2023 report, and I think this is also how we have to see these different formulations, and there has been a discussion on, on para five. As we see it, this is a development and this is similar to what Norway has explained of general principles that we have already developed in the context of the GGE laws in different reports and also in the guiding principles. And it's a way forward. And there seems to be a trend also in our discussions to just restate previous agreements that we have, but it's definitely also important to have a certain grade of development in these positions and in our formulations. And we all know especially when it comes to the 2023 formulation, how much we had to leave behind in these discussions also due to the shortage of time. So a development would only be natural in para five as a general principle in a possible preamble or part as it is, is an important formulation because it is also formulated in relation to how IHL applies. So this is the meaning and the essence of this para, and I think it's the right place. Talking about context-appropriate human control and judgment in box three has a different purpose. So it's very important to keep this in mind and also to see this difference and why it has been formulated in this way by you, Chair. So thanks again for having such an excellent text here. The second point that we wanted to make was this discussion about meaningful human control, and it very much relates to this as well. Obviously, meaningful human control, or also here context appropriate and meaningful human control and judgment, is something that needs to be explained. And many delegations have made this point that, as for now, it has not a lot of meaning. But it is our task here to fill it with meaning. We also have no problem to setting it out as a possible category of IHL, especially as Pakistan has explained so eloquently, as it is something that is needed for this specific technology and for these specific challenges that we are facing. This means that we have to do something actively. And box three, and this is something that we will make, this argument is something we will make later on again, box three is the place where we can feel this concept with meaning. And we will propose that there should be a chapeau where we define that context appropriate human control or meaningful human control or the longer formulation, whatever we agree. can be achieved through the following measures, and those could be the limitations and other measures that we set out in box three, which is not, as the US has argued, just filling up some word with content, but it should be a formulation that means that by doing or implementing the following measures or limitations, states can achieve meaningful human control. We will argue later why. It's just that through the overwhelming factors and speed that we will see with new technologies, we need to have certain factors that will allow a human to still be able to control or even employ judgment over the situation on the battlefield. And this is achieved through the measures in box three. Sorry for a little bit going forward, but this is very much related to the concept that is already spelled out here. Once again, the purpose here is a development of the previous agreements that we already had in the GGE on laws. Next point, and sorry for being so long, it's the issue of development and use. Pakistan has pointed out that if we are just talking about employment here and only about use, we are limiting ourselves to a specific subset of international humanitarian law. targeting law specifically. Development is not something new in IHL. It has been mentioned that it is a critical element of Additional Protocol I, more specifically on Article 36, but we have many conventions that are part of international humanitarian law which have a development aspect. I'm just talking about the PWC because it's very prominent currently in Geneva, which has a development aspect as well. The third point, and this has also already been made, the development part is a strong requirement in our discussions due to the nature of the weapons we are talking here. Due to the technology and due to the early formulating, the early shaping of the functioning and also about on the predictability, but also accountability aspects, and it has been mentioned about the bias that later on in the later stages of the life cycle of these weapon systems will hugely affect the functionality, but even the military usefulness of such a weapon. So it's crucial not only for the implementation of IHL, but also general for a proper functioning and for the safe deployment of such a weapon that the development phase is already considered. And the more paras we can include this, the better, and this is one of those paras very definitely need this formulation. Another point on meaningful human control, we note the strong reservation to this word, to the word meaningful. Once again, we believe that this is needed so that we don't have what has been called nominal human input. And we recall that this has been a concept that has been discussed in this room as something that we definitely need to address. And Germany has put it right, the one thing that we don't want to have is meaningless human control. And the reverse of meaningless human control, of course, is meaningful human control. The argument that this word has no meaning for now can easily be reversed by just saying we have to give this word meaning by spelling out. Obviously, we have a difficult text to negotiate in front of us, but one thing that we could also keep in mind is that we could also later on, in a later stage, have a technical annex and give examples, give best practices for meaningful human control in this part of the text and give it more life and give it more specificity. So there are many options to do this and just saying we don't know yet what it means is not a valid argument, at least from our side. Then maybe also quickly on the US suggestion on para five to put in the cardinal, we would not be sure because we prefer a broader characterization and I think in para 86 of the ICJ opinion it is talking about at least as far as I could research this until now, it is talking about the established principles and not about the cardinal principles. Established principles is interesting, so we could also use this, but principles and rules will probably be the broader formulation and the preferred one. I thank you.

Let me thank the distinguished representative of Austria for developing ideas and views with regard to the issue of human control and especially meaningful human control. Next on my list is the Philippines, to be followed by Costa Rica, Japan, Greece and South Africa. But first, let me pass the floor to the distinguished representative of the Philippines. You have the floor.

Thank you, Mr. Chair.

We see the text of box 2 as helpful.

And based on consensus language, especially the inclusion of the Martens clause in paragraph 4.

However, we are open to the suggestions either removing the first part or modifying it.

On paragraph 5, the Philippines prefer the term meaningful human control as an insurance against the minimal human input issue. We do understand the phrase context appropriate human control and judgment as a possible compromise language.

However, we feel that the current formulation is limiting and subjective.

We recommend opening paragraph five with human control and judgment in appropriate contexts. This ensures that human control and judgment is required in appropriate context and that we define the circumstances. under which the laws are used and its effects felt. Thank you, Mr. Chair.

I thank the distinguished representative of the Philippines for sharing the views of his delegation and making proposals. Next on my list is Costa Rica, to be followed by Japan, but let me first pass the floor to the distinguished representative of Costa Rica. You have the floor.

Thank you, Mr. Chair. Good morning, colleagues. We thank you for your work. And also, we thank all of you for being so involved in our discussions. Box 2 is a very good proposal, we believe, on paragraph 5. We believe it's important, as was mentioned by other delegations, to bring together the different criteria related to development and use. And also, Costa Rica reaffirms that meaningful human control is not something that can be negotiated. It is a fundamental principles to ensure that decisions are taken with responsibility, relevant levels of ethics, and with respect to IHL. Specifically, Mr. Chair, when we talk about questions of lethality, it is true that… We're involved in a theoretical debate, as mentioned by Pakistan, and the impact is quite meaningful. So we agree with Austria that we have to deepen our discussions of these concepts to have better clarity. The decision to attack cannot be reduced to an algorithm because wars is not a technical issue, but it's a deeply human issue. And the involvement of human beings related to autonomous systems in complex context, we have to make sure to apply the responsibility that is complete. If we leave it up to machines to decide on questions of life and death, then we will lose all controls. So this is why we are in favor of meaningful human control. Thank you.

I would like to thank the distinguished representative of Costa Rica for sharing the views of his delegation. Let me pass to distinguished representative of Japan to be followed by Greece and South Africa.

Thank you, Mr. Chairperson. And Japanese delegation considers the context appropriate human control and judgment, which is mentioned several times in this rolling text. requires clarification and common understanding of what it specifically refers to in each section or box. So my delegation understands context appropriate human control and judgment in this particular section, that is box two, to be the same interpretation as in the 2023 Lowe's GG report. paragraph 21c and the 2019 RoG report paragraph 17e. So based on this understanding, the contents of this para 5 of this box 2 are generally acceptable for my delegation. And the government of Japan is of the view that the presence or absence of human involvement is not a concept explicitly stated in the international humanitarian law per se. However, the lack of human involvement may result in the inability to operate as intended by the operator may cause unexpected damage to civilians. So my government places great importance on the operation within a responsible chain of human command and control and appropriate human judgment. So my delegation believes that it is important to continue to deepen discussions and form a common understanding on the nature of human involvement in those. Thank you very much, Chair.

Let me thank the distinguished representative of Japan for sharing the views of his delegation, next on my list is Greece to be followed by South Africa, and for now, South Africa is the last on my list. If you feel a need to respond to other delegations, and I see that there is clearly an appetite for that, and I welcome that because I think a few of you have been challenged by some of the interventions of previous delegations, so please raise your plaques and you will be added to the list. But let me now first pass the floor to the distinguished representative of South Africa. Madam, you have the floor.

Mr. President, I believe you had mentioned Greece before South Africa.

Thank you for reminding me. Let me first pass the floor to the distinguished representative of Greece, to be followed by South Africa.

My delegation supports the elements contained in box two and appreciates the efforts made to reflect key considerations related to the use of lethal autonomous weapon systems. We would like to focus our comments on paragraphs two and five of this box. As we have stated in previous meeting of this forum, the use of any weapon system in compliance with the provisions of international humanitarian law requires a degree of human control over its use on the battlefield. The compliance or non-compliance of weapons systems with IHL should be directly linked to the outcome of the attack and to whether the cardinal principles of IHL, as well as other applicable international law, including international customary law, have been respected before and during the attack. Therefore, in our view, it is essential to reflect the need for appropriate human control and judgment over the use of LOS, given that these systems must remain under a responsible chain of command and control. Regarding paragraph two, we would like to point out that IHL focuses on the principles that govern the conduct of armed conflict, such as distinction, proportionality, precautions in attack and military necessity. It does not depend on specific military technologies, most of which are dual use. In this context, we propose the deletion of the word military before technologies used to ensure a broader and more accurate reflection. Thank you, Mr. Chair.

Let me thank the distinguished representative of Greece for sharing the views of his delegation with regard to paragraphs two and five. Let me now pass the floor to the distinguished representative of South Africa. You have the floor.

Thank you very much, Mr. Chair, for giving me the floor again. Under this topic, I would like to highlight my delegation's view. We believe that human control is a critical element that should be considered under this box to ensure compliance. with international law, particularly IHL. However, we believe it is necessary to make a clear distinction between human control and human judgment, as these concepts imply different levels of responsibility and decision-making. My delegation is of the view that human control Particularly context appropriate human control seems to relate to the direct operation and engagement with the system, ensuring that human oversight is maintained throughout the targeting process. Human judgment, on the other hand, involves higher level decision making, determining when, how, and under what conditions laws may be employed. The responsibility for engaging a target and the responsibility for commanding its use are distinct and must be explicitly articulated to ensure accountability. Clarifying this distinction is essential for us, particularly in ensuring compliance with IHL principles of distinction, proportionality, and precautions in attack. We encourage discussions that explicitly recognise these roles to prevent ambiguity in the interpretation of human responsibility. Thank you.

Let me thank the distinguished representative of South Africa for sharing the views of her delegation. We now have a number of high contracting parties reacting to previous statements. I have Brazil, the USA, United Kingdom, and Germany, but let me first pass the floor to the distinguished representative of Brazil. So you have the floor.

Thank you, Mr. Chair, and good morning, everyone. My intervention, I don't want to belabor the points, just to react to some very interesting suggestions we heard last afternoon and morning. particular, my delegation wants to support the Japanese proposal on paragraph 2, which is to delete military. I believe it has also been supported by other delegations, in particular the US delegation. On paragraph 3, my delegation also supports the delegation of Colombia's proposal to add a more general reference to international law at the beginning of this paragraph. We can also support the additional ideas proposed by Mexico and supported by Portugal on specific references to international criminal law and human rights law. The proposal made by Sweden on paragraph 4 regarding the Martens Clause also seem extremely promising. And on paragraph 5, my delegation wants to thank the delegation of Pakistan for presenting cogent and powerful arguments in defense of the proposal made earlier by our delegation on the inclusion of development in use. I believe it has been extremely well presented, the case has been extremely well presented, and we subscribe to those arguments fully. We also would like to point out, as Austria, that the inclusion of this concept of human control in this section is absolutely necessary. This is a statement of an objective that should be inscribed into general principles of this instrument and of IHL as a whole. And the further development and operationalization of this concept would be done in box three. And also there is room, as suggested by Austria, to further develop even more this concept in technical annexes that can be added not only at the very beginning of the negotiation of this instrument, but further down the line. as technology evolves, it's to be expected that the challenges to ensuring human control and judgment will continue to evolve as well. So we shouldn't try to offer a final and coherent definition to this, to the operationalization of this concept, but affirming it is an absolute necessity in the view of our delegation. Thank you so much.

I would like to thank the distinguished representative of Brazil for reacting to some of the proposals made earlier and also in response to his remarks and those of the distinguished representative of Austria. It is interesting to see indeed that we again bring up the discussion we had, I think, at our first meeting last year with regard to possible measures, annexes and best practices. Yeah, we have not forgotten about it, clear. Now we have, of course, concrete text on the table, but those issues are, of course, something that we should keep in mind for future discussions. Next on my list is the USA, to be followed by the UK and Germany, but first the US. So you have the floor.

Thank you, Mr. Chairman. So we wanted to begin by thanking the delegation of Pakistan for setting the tone for our debate this morning with what we thought was a very, very thoughtful intervention. I think at this stage, we're really trying to achieve a lot more granularity on what the purposes of the instrument are, what problems we're trying to solve with this instrument, and we think that this has set the tone for this morning's conversation in a way that was quite helpful. We did want to engage on several of the arguments that their delegation made, and arguments that have been picked up by a number of others in the room this morning. So first, with respect to development, the delegation of Pakistan, I think, made a convincing argument that some sort of framing that's based on Article 36 principles in a way that's acceptable to parties like the United States who are not themselves parties to Additional Protocol I is a fruitful way to approach the need to articulate a legal principle related to development. I think everyone actually agrees with it. And Chair, I think your own text recognizes that by including exactly that framing in box four, Article I. And so I know we're very focused on box two at this point, but as many delegations have pointed out, it's important to sort of see how this text is developing as we frame provisions in certain boxes that relate to provisions in other boxes. So we would just point out that this is something that's actually incorporated in your rolling text in a place that we think is actually more appropriate in box four. Second, with respect to the debate that we've been having on paragraph five and the reference to context appropriate human control and judgment, I think for some delegations meaningful control, I think the starting point for our colleague from the delegation of Pakistan was to recall that paragraph 21 of the 2023 GGE report does refer to control. And we mentioned this yesterday. I can just read it again today. It indicates that control with regard to weapon systems based on emerging technologies in the area of laws is needed to uphold compliance with international law, in particular IHL, including the principles and requirements of distinction, proportionality, and precautions in attack. So this paragraph, as anyone who was in the room in 2023 will remember, was the subject of extensive negotiation. And we essentially had the same debate that week and that evening that we're having today. That's not to say that we don't welcome continuing to have this dialogue, but just to point out that we're not drawing on a blank slate. And as we indicated yesterday, we're ready to use this formulation again in this paragraph. I think the central argument that the delegation of Pakistan made as to why we need at this point to innovate and to go beyond what we've previously agreed is the view that it was not previously necessary to spell out that human control is required because IHL has always assumed it. I think that's fairly close to the way the point was made. And based on that, the delegation of Pakistan has challenged us to ask whether IHL requires human control Or can machines exercise control over the critical functions of a weapon? And I think again, that's the way the delegation of Pakistan framed the question and the challenge before us. I think our delegation believes that the answer to this question is pretty clear. As we've been discussing for years, the ability to identify, select, and engage targets which are undoubtedly critical functions, we've already identified them as among the essential characteristics of laws, are functions that are now in weapons that almost every delegation possesses, controlled by autonomous technologies. So we have at least the beginning of an answer to the question that suggests that human control is not always necessary I think we've agreed to that concept that's been a central assumption in our in our discussions chair we've listened carefully again this morning to the debate surrounding this paragraph um we continue to hear multiple inconsistent divergent explanations of what this means to individual delegations We have heard the suggestion this morning again by the delegations of Austria and Brazil that we should agree to including the reference now and then develop an annex to explain what it means later. I think from our delegation's perspective, this is the wrong order. We're very open to developing language in the rolling text, in an annex. that identifies the rules, the regulations, the practices, the measures that animate what delegations think they mean by meaningful human control or by context appropriate human control. But if we achieve consensus on such an annex, then for us it really does beg the question why we need the label. So from our perspective, we still aren't convinced that this is an approach that's going to achieve the clarity on the rules that apply to states in relation to laws or on the measures necessary to comply with those rules. Thank you.

Let me thank the distinguished representative of the US for responding to previous remarks made by different delegations. Next on my list is the United Kingdom, to be followed by Germany, Australia and Article 36. But let me first pass the floor to the distinguished representative of the UK. You have the floor.

Thank you, Mr. Chair. Mr. Chair, we've also been reflecting on the useful discussions this morning and thought that it's been really useful to hear the different views around the room in relation to what this paragraph five is actually getting at. In particular, we thank the distinguished delegate of Pakistan for his intervention, which was really got into the core of the issues I think we're here to discuss today. Equally, I think it was also useful that the delegate from Austria has reminded us of what we're here doing and that we are effectively negotiating something. We're looking for a form of words, a form of text to capture something. And so that is useful as we start to try and hone in on these concepts that we're hearing around the room as to how we actually then capture this on the page. We are hearing that the cognitive element that these particular systems demonstrate raises the question of what is the role of the human. This is the difference from the conversations that have been had under the auspices of the CCW in relation to all the other protocols and the other conversations that have been held in these rooms. We recognise that these labels of human control, context appropriate human control, meaningful human control are labels. And we've heard a number of delegates this morning saying that they each understand these labels to mean a particular thing. What I haven't heard necessarily is a convergence exactly on what those labels are meaning. And so in our view, this may well be part of the problem. And I think I've just heard that from my colleague from the US just before this intervention. So how do we then develop text that avoids this problem of a label that can then be taken away to mean different things to different states? Our suggestion is that if we're more assertive in the phrasing, if we actually think about whether or not we need to adopt a label in the phrasing that we use, or whether we can actually just say what we mean. We've used control in the various documents that we've concluded as a part over the years as part of this group. Control as a verb, I think, is an understood term. Control in relation to the system is an understood term. Equally, we understand we've used context of operations as well. So I think if we went back to paragraph five and actually looked at that in a way that takes out some of those labels and reflected on this issue of actually, I think we understand what we're getting at, but we need to make sure that we don't have divergent interpretations. We can come out with a slightly more straightforward approach to the concepts that we've got there. So I think we're looking at control and human judgment and context and whether we can find a form of words that will capture that so that we can keep meeting this requirement across the room to have something that captures the difference between previous systems and these systems that have a cognitive element. Thank you, Mr. Chair.

I would like to thank the distinguished representative of the UK for sharing her views on this. And yes, indeed, I think with this we are at the heart of the matter. And It's always a temptation to look for constructive ambiguity, but I don't think that will work here. We need to have clarity on it and there's time to continue the discussion, so we will continue doing that. We will do it today and we will continue this discussion also after today, I think, because this is box two and is also very much related, as some of you already said, with box three and four. And three and four, we try to have at least two readings of this week. But let's see. Next on my list is Germany, to be followed by Australia and Article 36. But let me first pass the floor to the distinguished representative of Germany, Madam Wefer.

Mr. President, with regard to paragraph 2, Germany supports Japan's suggestion to delete military before technology' so that it reads applicable independent of the technology used'.

I would like to thank the distinguished representative of Germany for responding to a previous proposal. Next on my list is Australia, to be followed by Article 36 and Austria. But first, the distinguished representative of Australia. Madam, you have the floor.

Thank you, Chair. Australia would also like to respond to some of the points raised on the floor regarding human control. We have listened with interest to the concerns raised by various delegations regarding human control, its relationship to existing IHL, and the emergence of laws forcing us to ask questions not previously asked regarding the adequacy of existing international law frameworks. Chair, we wish to make a few observations in response. Humans must at all times remain accountable in accordance with applicable international law for decisions on the use of force. And the use of autonomous weapon systems does not provide the basis for excluding legal responsibility. By way of an example, a person who activates an autonomous weapon system with the knowledge that it will strike military objectives and civilians or civilian objects without distinction could be held individually responsible under international criminal law. To reiterate, Chair, established legal principles of accountability would apply in relation to the use of autonomous weapon systems. For example, IHL requires parties to a conflict to take necessary measures to prevent and suppress serious violations of IHL, and serious violations of IHL involving autonomous weapons systems may give rise to individual criminal responsibility under international law. The discussions this morning have been illuminating and productive, and we look forward to continuing these with you all throughout the course of the week. Thank you.

Let me thank the distinguished representative of Australia for sharing her views and indeed, uh, as was done, uh, already early on discussion, also linking, uh, our discussion we have now on box two with the discussion we still are, are going to have, uh, on box five with regard to, uh, responsibility and accountability. next on my list is article thirty six to be followed by austria and the russian federation. let me now first pass the floor to the representative of article thirty six. you have the floor.

Mr. president thank you delegates for the thoughtful discussion this morning. i just wanted to comment in broad terms on paragraph five and i guess returning to themes that a number of delegations have brought up. I think for us, this paragraph introduces a key sort of test of the text that you've circulated, a description in a way of the form of judgment and control that's needed. And as such, it's really tying to my mind the challenges, the problems that autonomous weapons pose. the obligations of existing law and recognising that the reason why we're having these discussions is because we are collectively recognising that there's a need to find some connective tissue between the problems that we see these systems presenting and our commitments to established legal structures. So this paragraph is in one way a bridge between the challenges of the weapon systems and the law, but also perhaps in the structure of the text itself, it's a bridge into into the text that follows, it does, to my mind, lack the positive moral implication of meaningful human control. But in our understanding of that phrase, it was always intended to be understood as being context dependent and requiring context dependent determinations by humans. And I think the text that you have here does bring that out. judgment, of course, brings out the sense of deliberative human engagement that I think we recognise as being core to the human element, but also a fundamental building block of the law. The law is structured around moments of human deliberative engagement. But control, as a number of delegations have mentioned, is necessary in order for that judgment to come to bear and to be brought to bear on the technologies that we are talking about here. In terms of how this should be interpreted, uh, I mean, it feels, it feels to me that in a way this is, this is leading us into box three and it's in box three that we start to get a more refined, uh, articulation of how the context appropriate human control and judgment should be understood. It's understood in the context of the subsequent tests that those paragraphs set out. So I don't see this as being sort of approached separately in some separate annex that it needs to be defined, but rather it's defined through the rules and obligations that we bring in in the subsequent boxes of the text. It's context appropriate after all, so I think it wouldn't be surprising if those tests, those subsequent tests, are also somewhat open-ended. We can't have a requirement for context appropriate judgment and control and then expect to stipulate the terms of that so precisely that there is no space for the context appropriate flexibility that is at the heart of the matter. And I think when you get into box three, the key tests that you've set out there go a long way to helping us to understand how that term should be understood. So I suppose this is really a suggestion that we understand this phrase in the context of the text as a whole. certainly don't see the need to then imagine that we are requiring control at every single granular level of a weapon system functioning because that's clearly not the case in practice and I don't think anybody would actually interpret control to require that. I think that's, yeah, through its subsequent definition and articulation in box three and box four, that's where I think we can see this phrase unpacked and made tractable in a sort of legal sense. Thanks, Chair. I.

Would like to thank the distinguished representative of Article 36 for sharing his views on the discussion so far and also on discussions still to be followed with regard to box number three. Next on my list is Austria, to be followed by the Russian Federation and Mauritania, but let me first pass the floor to the distinguished representative of Austria. Sir, you have the floor.

I just want to thank all the delegations. I think we are having an excellent discussion this morning and we are getting deeper to the core of the issue, and it is one of the most crucial issues here, so I really appreciate the discussion that we are And I also want to support what you have mentioned, I mean, what we need to avoid is constructive ambiguity. I always felt that this, we all know it's, it's, it's a reality of diplomatic life, but I always feel like it's like burdening future generations of diplomats with, with something that we were not able to resolve. And this was maybe also the spirit in which we suggested to have a technical annex. I don't think that we have to first negotiate a technical annex, but that we first have a clear agreement what we mean, what these concepts mean for each of us, and then we try to explain it. But this is not something that specifically we really require. It's more something that could help. future generations of people who try to look into what we have formulated here and don't -- to avoid that they have to read it like a holy text or something, that we give some explanations what we actually meant. And this is also common practice with international treaties. Usually there is an explanatory memorandum or something like this, at least in other fields outside of disarmament. It is something that is common practice where we have less disagreement. But in the CCW context, as you all know, a technical annex is also something that has been already done. But it should be the afterthought and not something that we have to agree first. Then I wanted to pick up the argument that human control would kind of indicate that it is -- or how we formulate it here, that it would indicate that human control is always necessary. I think we have kind of gone away from that. I mean, we have, of course, the formulation on control without the human in the 2023 report. But what we are seeing here on the screen already is a major concession because we are talking about context appropriate human control. The term has found large agreement as far as I can see with a few exceptions here in this room. But this kind of shadows that this is something that is kind of a compromise already because it is taking away something from the human control and already indicates that there are situation, there are different layers, different -- a word that we used in the past where there is a certain kind of spectrum of how this human control could be implemented and what kind of situations it is more needed and in what kind of situations it is less needed. So this is something that already takes a huge step on the one side into a compromise, but on the other side, it is also something that kind of weakens the word of human control. And we have to stress this once again, this is a concession also from our side to be able to consider this. So I think we all have understood this, and this is why there is not much discussion about the context appropriate formulation, but we should be aware that this is a step into one direction that is a huge step for many delegations. I just wanted to stress this here again. Then, and I'm sorry for this, on the meaningful. I think what the meaningful does is addressing a problem. And this is a problem that we actually have to discuss. It's the one about nominal human input and about the situations that then also lead to something known as automation bias and that just undermines the rest of the text or, in other words, creates a major loophole. So I think it is absolutely necessary, even if there are different understandings about meaningful or if this is a word that some delegations cannot accept that we have a discussion about this problem and how to resolve it, because otherwise we are creating, as I mentioned before, we are creating an instrument that is full of holes and that can be easily undermined by state practice and by different tactics or methods. So this is something that we need to continue to discuss and I hope that we can do this also in the later phases of the text. I thank you.

I would like to thank the distinguished representative of Austria for further elaborating his positions. I have now the Russian Federation to be followed by Mauritania, but let me first pass the floor to the distinguished representative of the Russian Federation.

You have the floor.

Thank you, Mr. Chairperson. Our delegation would like briefly to come back on the proposals by a number of delegations connected with the references to international humanitarian law in two and three which it was suggested that it's not enough to have just a reference to IHL there needs to be one to international human rights law and international criminal law this issue is not new for the GGE. We've been discussing this for over 10 years here. But despite that, no result has been obtained on the basis of that because delegations do not have consensus on this matter. And we see that as things stand, it's going to be difficult to get that consensus. So, as we see it, we need to focus particularly this stage on the things that bring us together, where positions align among the states. And alignment of positions can be seen, we believe, as regards the reference to international humanitarian law. So we believe we should focus on that, since getting consensus on international human rights law or international criminal law is going to be difficult. Thank you.

Mr. President, let me thank the distinguished representative of the Russian Federation for giving his reaction to proposals made previously by other delegations. I have now on my list Mauritania. Let me pass the floor to the distinguished representative of Mauritania. You have the floor.

I would like to congratulate the Chairperson and start, as I have said, by congratulating the Chairperson and his staff for the guidance given to this conference. It is a great honour for me and a genuine pleasure to be here with you. at this meeting to discuss such a pressing humanitarian issue. We're looking at something assimilated to killer robots. There is no definition, no shared definition at this stage, nor any binding legal framework internationally and this gives rise to major concern, particularly if we look at the speed of technological developments. This concern is an ethical one, as well as being a legal and a security concern. In terms of international law, the issue is that autonomous weapons systems, which are not under human control, represent a danger to all of humanity, even though some point to certain definitions. They are a danger, as I say, to all of humanity because these weapons can lead to unpredictable wars, given the speed and in particular the lack of command and control, the lack of a command and control chain. Now, science without conscience is a danger. It would be wise, therefore, to have terms of reference if there is no shared definition. We need to know the implications and the position of the human being in the operation of these weapons systems. The presence of a human being is mandatory in order to make these systems operate. Thus, we see the limitations of this use, because we see the involvement of the individual in the weapons system, which makes itself no distinctions. Thank you.

Let me thank the distinguished representative of Mauritania for sharing the views of his delegation with the group. We are at the end of my list with regard to requests for the floor, so I am just looking around once more to see if anybody would still like to take the floor and react, respond to the things that have been said today and yesterday. And if that's not the case, I am now tempted to, well tempted, I'm right, yes, to try at least to give a bit of a summing up where we stand with regard to box number two. And I think I could say it's a kind of a mixed box because some of the language that's clearly More or less okay, and there are some further improvements, and I think we have, my team will be able to… to do some work on that. We have a number of helpful suggestions, for instance, with regard to the Martens clause, and we will look into that. Also, with regard to development and use, and with regard to the phrasing of the the last part of para two on applicable independent of the military technology to use. On four, there were, sorry, there were also some proposals. We will look into how we can take some of them on board to bring a greater, let's say, convergence in the room. And then, of course, the really difficult issue was para number five. It has been said there's a link with section or box number three and four, and we will discuss them shortly. There's also, as some of you said, there's a link with the text in box number five with regard to responsibility and accountability. And of course, as I said before, this is really one of the most essential things we have to discuss this week, the issue of human control and judgment, how human control and judgment relate to each other, what, as was also explained by some of the delegations, what is the link with context appropriate, all issues that really deserve further discussion and that will come because this is also something we will meet again when we discuss the next boxes. References have been made to texts from 2019, 2023, and I think that's very relevant. On the other hand, we are not a kind of an historical society. So it's not just finding effect in the future that is then the highest wisdom we have for the present effect in the future, which is then guiding us in the present and also for the future. It has been pointed out also by some of the delegations. We have, of course, a reference also in our mandate to the work done, previous work of the group. So we should truly take it into account, but we should also, again, discuss it. in the light of also ongoing developments. The world does look differently from 2019, I dare say. So clearly, especially with regard to paragraph five of this box, we have more work to do. I would like to thank delegations for coming up with proposals, and we will see how we can work on the basis of them. We will also reach out during the week to delegations to further discuss that and to indeed further understand their positions and their proposals made. So this is with regard to box number two. Let me now then turn to the third and fourth boxes of the rolling text. Already quite a number of you made references to both of them, so let me also do that here. Box number three and four contain elements for possible prohibitions and regulations in line with what some call here, or quite a number of you call, the two-tiered approach. For many, these sections are the core of the rolling text and, as was said also by a number of you, Paragraph five of box number two was in a way an introduction to box number three and four. The informal consultations that we have held during the intersessional period, including with observers, gave us many suggestions with regard to further evolving and strengthening the text, including by adding the word development to use and standardizing it to development and use as per accepted language in disarmament treaties. Some of you also have pointed out that not in all cases they would like to see that because of the context of the para. We have to take that into account, I think. Furthermore, a number of delegations highlighted certain questions of a truly conceptual and terminological nature. Let me recall them for you and for the purpose of our discussion for the rest of this morning and this afternoon. The main outstanding issues include the formulation of prohibitions. and more in particular the way they are formulated also. So is the current formulation, it is prohibited to the most suitable formulation to reach consensus on text elements regarding possible prohibitions, or do we need to look at other formulations like must or not be used, must not be used, sorry, as it is already contained in the Greek language of the group search as in the 2023 report, power 21 Bravo. Another outstanding issue is, and there was already reference made to that, the list of limitations. Limiting the types of targets to military objectives by nature added to address the aspect of laws. This has met with some objections, noting that the military nature of objects is not fixed and can change depending on the circumstances of combat. Questions were also raised as to why the number of targets should be limited. Delegations have stated that mission parameters of laws should be tied to the critical or core functions of laws, read selecting and engaging targets. In light of the extensive feedback we received on this specific part, it will be appropriate to consider whether or not we should frame or categorize this section on regulations in a different way. For instance, taking existing principles of IHL as a starting point. More specific terms like predictability and reliability can then be explicitly linked to concepts like proportionality, distinction, and precautions in attack. Then, of course, the issue of human control. Discussions have focused on human control, appropriate levels, and retention of control throughout the life cycle of a weapon. Some delegations have also argued for clear explanation as to what is prohibited or restricted, while some stated that this remains a national prerogative. Earlier today already, reference was made to nominal human input. The concept of nominal human input was reintroduced in this section through the need to ensure that moral and legal consideration is given by human. While there is broad agreement on the need for legal consideration, concerns remain over putting legal and moral judgment on the same level. With regard to box number four, we have received a number of proposals with regard to further evolving the text. For instance, regarding the need to be clear about measures to address data bias and automation bias. At the same time, a number of questions and concerns were expressed with regard to the language now in the text. General questions were raised also over the use of operational verbs, that is, should or must in the Paris in this section, bearing in mind the possible need to differentiate between existing or possible new legal obligations and possible recommendations. Let me also remind the group of what we are doing is without prejudice to the legal character of any elements we are developing. A number of delegations also emphasised that technological innovation should not be hampered for any of the proposed regulations. And that's something that was mentioned already in para J of the guiding principles in 2019. Let me now invite delegations to take the floor and to provide their views, but not only views, I would also appreciate, of course, to receive possible solutions for still unresolved issues. as you have done also during the previous discussion on the boxes one and two. Let me also highlight that there is no need to be reluctant to open the debate because, as you have seen, you can always come back for a second and even a third time if you would like to do so. I would even encourage that, to have an interactive discussion and to have a better feel of how different delegations look at proposals made by colleagues in the room. Having said that, let me now open the floor for the discussion and perhaps you could show again box number three, but you are free to reflect on both three and four, or if you would prefer that to start with three and come back for four. Let me thank the distinguished representative of Ireland for breaking the ice. Anyhow, I'm not going to send you off to lunch yet. We have almost an hour to go and we really have to make the best use of our time. But again, many thanks to the Irish delegation and let me pass the floor to you.

Thank you, Mr. Chair, as well for your your summary highlighting the main issues in this section, and I'll keep my remarks just to section three for the moment. Section three includes many essential elements in ensuring meaningful human control and judgment necessary to comply with IHL in line with paragraph five of the previous box. We welcome paragraphs one through four, which reflect fundamental IHL considerations. In line with other delegations, and as we expressed yesterday as well, we would suggest to use the term develop and use laws in paragraphs one, two, four, and five of this section for the reasons highlighted by many yesterday and today. Again, we welcome the references to context appropriate human control and judgment and would prefer to see meaningful before human. And just want to thank the delegation of Pakistan, among others this morning, for very effectively outlining why including both development and use and context appropriate human control and judgment are essential to our work. On paragraph 6a, we think this should be pulled forward into the prohibition section through an additional paragraph that would read, It is prohibited in all circumstances to develop and use laws the effects of which are not predictable, reliable, traceable, and explainable. The reasoning for this is that, as required by IHL, users of weapons must be able to, with a reasonable degree of certainty, predict the effects of that weapon. In the context of laws, this requires including the concepts of predictability, reliability, traceability, and explainability. And we heard some questions in the intercessional consultations about what is meant by some of these concepts, in particular, traceability and explainability. So I will just provide a couple of clarifications on these two concepts. Traceability is the ability to track and document a system's decision-making process, development, and deployment, ensuring there's a record of how the system was designed, trained, and tested, and how it made specific targeting decisions. This is crucial for ensuring accountability and to allow investigations in case of errors or unlawful actions. Explainability is the degree to which a human can understand how and why a law has made a particular decision. We understand this is particularly difficult due to the nature of machine learning and complex algorithms, but it is necessary requirement for users to understand the system's limitations and potential biases. For paragraph six, we believe the chapeau should be amended to read, To ensure that all other laws only operate with context appropriate human control and judgment and to uphold compliance with IHL, parties to a conflict need to. And this is again in order to elevate the centrality of the human control element. For 6, we welcome the inclusion of moral and legal consideration given by human as a way, as you described, to address the nominal human input issue. And again, we heard during the intercessional period that some delegations take issue with morality included here due to its subjective nature. And perhaps we could find some language regarding ethical considerations. But as mentioned by Austria, the nominal human input loophole can be closed through using the term meaningful, and this is something we suggested in box one as well. On paragraph 6c, we very much welcome this important section on limits, and we support the sub-paragraphs contained. On Roman numeral five, we suggest the deletion of, for instance, by considering to restrict the use of laws against in order to avoid any confusion and to have a clear limitation on military targets. And I'll leave it there for now. Thank you, Chair.

Let me thank the distinguished representative of Ireland for his remarks and suggestions with regard to the text proposals. Next on my list is Switzerland. Let me pass the floor to the distinguished representative of Switzerland. You have the floor.

Thank you, Mr. Chair. At this stage, I would like to share a few general considerations and some specific comments on the individual paragraphs of box three. And as you offered, we'd like to come back on other aspects at a later stage. Section three is a crucial part of our discussion and contains pragmatic and implementable measures to preserve control needed in order to ensure compliance with international humanitarian law and addresses ethical, legal, and operational challenges posed by autonomous weapon systems. Switzerland would push for a forward-looking approach, emphasizing that while past agreements are useful references, they are not sufficient at this stage and on their own. Rather, our understanding is that the GG has in its mandate engaging now with essential questions of prohibitions, restrictions and obligations, moving from principles to actionable commitments that can form the basis of a future instrument. The general principles and previous GGE reports already outline broad understandings and commitments such as compliance with IHL, the necessity of human involvement in the use of force. However, to fulfill the 2026 mandate, the GGE must now work towards language that defines clear obligations. and prohibitions. The GGE is expected to produce substantive outcomes, not just discussions. If the remainder of the GGE's time is spent restating previous conclusions, it would fail to fulfill this purpose. Previously agreed language is not the end point. For instance, if in the past we reaffirmed the importance of human control, we now define what levels of human control are required and where prohibitions apply. If we reaffirmed in the past the applicability of IHL, we should clarify what behaviors of autonomous weapons or functions are incompatible with IHL and must be prohibited or restricted. While we see the need for continuity, the purpose of this GGE is not to restate past principle, but to translate them into elements in view of concrete normative and legal commitments. Instead of merely repeating previous GGE reports, we should be building on them to clarify what they mean in terms of practical obligations and legal restrictions, which we which is why we find box three so important as it is stated there on the screen. Some general points on how specific and exhaustive we must be. Already in the informals we have heard that there's been reflections on expanding the list of prohibitions and And there's a question where to draw the balance between the continued applicability and validity of general rules of IHL, which continue to apply to all means and methods of warfare, including autonomous weapons, and adapting the existing rules specifically to refer to autonomous weapon systems. We see a risk that we basically would need to copy and adapt all IHL rules, especially those on the conduct of hostilities, which would probably be too exhaustive and raise complications. Furthermore, the more rules we adapt, it could also raise questions regarding the continued validity of other existing rules that we have not adapted. So we think there is a fine balance to find. On terminology, When referring to IHL, we generally would prefer using the term rules instead of requirements. This is because IHL is a set of rules and principles that seek for humanitarian reasons to limit the effects of armed conflict. On the individual paragraphs, and here for the sake of brevity, I would for now limit myself to paragraphs one to five. and then come back later on the remaining paragraphs at a later stage. Paragraphs one to four are fundamental IHL considerations and guiding principles in previous reports. We don't see them as a mere statement of fact, but for us they are really commitments that States have to sign up to and comply with. Like others, we would generally support the development and use of lethal autonomous weapons. Paragraph 2, we support language referring to inherently indiscriminate or indiscriminate by nature as a short form. While there is some that say that it's not treaty language used widely, we believe it is a commonly used term that we should also be able to use here. If a definition would be needed, we would have to expand it considerably and we could come back on that at a later stage. Paragraph three, We support the language as it stands, which seems to us rather standard and widely used as it combines, for instance, Articles 51 and 52 of Additional Protocol I and also follows language used in the CCW, for instance, in Amended Protocol II or Protocol III. Paragraphs four and, in particular, five are key paragraphs that we support as they are. Thank you very much, and we will come back on the remainder at a later stage.

Let me thank the distinguished representative of Switzerland for sharing the views of his delegation with regard to box three, in particular the first five paragraphs. I am looking at my list. So far, no other requests for the floor. Still have about half an hour. So, I would like to also share some initial views with regard to box number three. I recognize the US and to be followed by Sweden. Let me first pass the floor to the distinguished representative of the US to be followed by Sweden. And thank you for picking up this.

Thank you, Mr. Chair, for giving us the floor. We just wanted to start off with a few comments on section three. I think one question that my delegation was trying to work through is, you know, what is the difference between box three and box four? Now, I think there could be, there have been different interpretations of the two tier approach and it hasn't precisely reached consensus in the past GGE discussions. I think our understanding is that box three relates to prohibitions and restrictions. And so that would be our title if we could reach consensus on titles at some point. And then box four, in our view, really relates more to implementation measures. So that's, I think, how we would see the structure of the boxes. But of course, we welcome other viewpoints and clarifications on that aspect. Now diving into the maybe the first couple paragraphs, you know, I think one issue we wanted to point out first is what we saw as a redundancy between the first paragraph and the second paragraph. And so, you know, for example, The prohibition of using weapons which are inherently indiscriminate, I think our Swiss colleagues spoke very well on that point, as this is a common concept in IHL, which is widely used, and we support that idea as a core IHL prohibition. We also support including it in this set of elements of an instrument. But for us, that idea of inherently indiscriminate weapons is also what is addressed in paragraph one. For us, these weapons are weapons that are capable of being employed in compliance with IHL, in particular, the principles and requirements of distinction and proportionality. So we would read actually paragraph one and paragraph two to be duplicative. And then for us, that raises a legal concern. If you have the same rule, but it's articulated in different ways in the same instrument, that actually implies that there may be different rules. And so we would therefore recommend merging boxes or paragraphs one and paragraph two. And I think that approach would also, I think, align well with some of the past foundational elements that we found consensus on in the past. I think I'll draw colleagues' attention to the 2019 GGE report, paragraph 17H, as well as paragraph 21B of the 2023 GGE report. So more specifically, our suggestion is to omit the first paragraph and then add to the second paragraph, or which are otherwise incapable of being employed in compliance with IHL. And so I think we think this would track well with previous agreements. It would establish this foundational principle that we're focused on weapons which are by nature intrinsically capable of use in compliance with IHL. And then we would avoid legal complications with kind of redundancy or suggesting that the prohibition on inherently indiscriminate weapons is somehow different from the language of paragraph one. I also wanted to note here that for us, the rule regarding the taking of precautions and the planning and conducting of attacks, that requirement, that rule, is not a basis for a per se prohibition on the use of a weapon. For us, the rule on precautions in attack, in planning and conducting attacks, is a very contextual, fact-specific rule. There's, I think, a well-accepted interpretation and practice about what is considered a feasible precaution. And it depends on military and humanitarian considerations specific to circumstances. And in our understanding, this is not generalizable to a per se prohibition on a weapon under the law of weapons. So therefore, that's why we've omitted that in the formulation that we're proposing with regard to paragraph 2. I think the next point that we wanted to make was to propose an additional paragraph. you know, we saw this idea in paragraphs one and two, which is reflecting the core prohibitions on weapons law that we found in past GGE reports and consensus that we're going to try and evolve and build upon in a progressive way in the rest of the instrument, you know, but we also thought it was helpful, it would be helpful to add a paragraph articulating the key requirements and principles of targeting law. that the GGE has already found consensus on and that are of particular relevance in the context of laws. And so our specific proposal for a paragraph would be, "Laws may only be used in accordance with the principles and requirements of distinction, proportionality, and precautions in attack, in particular:" And so this idea would be to sort of complement the very nice sort of paragraph articulating core elements of the weapon law with similar elements with regard to law of targeting. And then it could then set a foundation for the following paragraphs, which could present more specific articulations of distinction and proportionality and precautions and attack. So I'll stop here with just comments on these introductory paragraphs, but really thank the chair and colleagues for their attention.

Let me thank the distinguished representative of the US for sharing comments on the first part of the text of box three. I have now Sweden, Argentina and Campaign to Stop Killer Robots. Let me first pass to the distinguished representative of Sweden. You have the floor.

Thank you, Chair. So Sweden supports the majority of box three. with a couple of exceptions that I will return to. We believe that this box builds upon existing IHL, has a clear loss context and balances concerns and opportunities in a good way. It contains elements that in concrete ways may serve to ensure meaningful human control and that laws can be used in line with international law, in particular IHL. We do, however, would like to see some smaller modifications. Instead of using principles and requirements of distinction, proportionality and precaution, we would prefer rules and principles. This concerns the whole rolling text, by the way. We also suggest removing in all circumstances from para three. In para six, Sweden would like to change need to to should. In B, we would like to strike out moral. In C, we would change including through to inter alia through or similar exemplifying formulation that we can find. And we would also remove three, four and five. We also note that the first part of para C already mentions limitations of type of targets, duration, geographical scope and scale of the operations in a manner that corresponds to already agreed language. But that's all for, from our part at this point. Thank you. Thank you, Chair.

I thank the distinguished representative of Sweden for making a number of concrete proposals, also fruitful thoughts for other delegations to think about. Next on my list is Argentina. You have the floor, sir.

Thank you, Chairperson. We believe that this proposal is largely sustainable. We would like to make a few particular comments. On paragraph two, we agree with the emphasis on prohibiting indiscriminate use, so like in the other protocol. In three, strengthening the concept of human control, we believe that we could broaden it by mentioning the necessary precautions that must be taken in densely populated areas where it's difficult to identify objectives. We recognise that this reference is included in paragraph six, but we believe it would be useful here as well. In paragraph five, the term context appropriate human control and judgement, as we see it, is part of the consensus solution. We can accept that. But we think it's a bit ambiguous here. We propose defining the term on the basis of a) characteristics of the system, whether it's designed to allow human judgment and control, whether it's reliable in decision-making, and b) the characteristics of the operational system, whether it's predictable and so forth. We also recognise that part of the issues are present in other paragraphs. In paragraph 6, D, we suggest changing moral for eth-- in B, exchanging moral for ethical. That's more in line with the other wording. And in six, C, that's more relevant. Five, we have reservations regarding the new vocabulary and the military objectives by nature, that concept. We understand that the term is present in Additional Protocol I, but we believe it is not the object of universal consensus and there are disagreements on whether military objectives by nature are an absolute category. So we would prefer not to have that. Thank you.

We thank the distinguished representative of Argentina. for commenting on the text and providing suggestions. Next on my list is the campaign Stop Killer Robots, to be followed by Brazil and Israel. But let me first pass the floor to the representative of the campaign to stop killer robots. You have the floor.

Thank you, Chair. The Stop Killer Robots campaign welcomes the attempt to include anti-personnel systems in box three on restricting use to objects that are military objectives by nature and urges States to provide more explicit rules on anti-personnel systems to effectively address core legal and ethical challenges, including around bias. Stop Killer Robots is aligned with the position of the International Committee of the Red Cross on this issue. From a technical perspective, distinguishing valid human targets is immensely more difficult than identifying military objects by nature, such as tanks, munitions, or military facilities. This is because there is no stable and universal combat target profile. This is a contextual human judgment to make, which makes it impossible to build a software system that can accurately distinguish combatants from civilians. There is also the further problem of recognizing combatants who are hors de combat, injured or surrendering. The risk of IHL violations in this context of the use of autonomous anti-personnel weapon systems is high, as is the risk of friendly military personnel who may be present where such weapon systems are used. Moreover, the intentional killing of a human being requires legal and moral justifications. No machine, computer or algorithm is capable of recognizing a human as a human being, nor can it respect humans as inherent bearers of rights and dignity, understand what it means to be in a state of war, much less what it means to have or to end a human life. Decisions to end human life must be made by humans in order to be morally justifiable. The current paragraph of the text in box three on limiting the types of targets, durations, geographical scope and scale of the operation of autonomous weapons systems should be extended to explicitly prohibit targeting of personnel, and we urge States to consider making this addition. Thank you, Chair.

Let me thank the representative of Stop Killer Robots for sharing her views. Next on my list is Brazil, to be followed by Israel, but I will first pass the floor to the distinguished representative of Brazil.

Thank you, Mr. Chairman. Like other delegations, Brazil believes that this box is the very core of this document and we regard it as a concrete reflection of the long-standing aspirations of delegations to define a two-tier approach for the regulation of autonomous weapons systems. Regarding the areas in which the language could be fine-tuned, my delegation fully concurs with the positions that were expressed this morning and in the informal consultations regarding the need to insert development and use in paragraphs 1 to 5. Like the delegation of Ireland and other delegations in the informal consultations, Brazil believes that the chapeau of paragraph 6 should be revised in order to clarify that the requirements set forth in its subparagraphs are what defines the concept of context appropriate human judgment and control, or as many delegations have suggested, context appropriate meaningful human judgment and control. In this regard, my delegation believes that subparagraph 6 is a critical element of this concept. as the lack of predictability, reliability, traceability, and explainability would render moot any possibility of human control. As for subparagraph 6c, small Roman 5, many delegations expressed concerns during informal consultations regarding the lack of clarity about the limitation of the number of engagements that autonomous weapon systems can undertake. My delegation believes that this formulation aims at avoiding the possibility of laws entering into an uncontrollable loop. Therefore, in order to reinforce the clarity of this language, my delegation would suggest the alternative formulation, open quote, limiting as appropriate the number and timeframe of engagements that laws can undertake prior to seeking additional human operator input." End quote. I can, of course, send this suggestion to the Secretariat. And as for subparagraph 6c, small Roman six, we concur with the observations made by some delegations in the informals about the need to clarify that this language refers to the critical functions of laws. In this regard, my delegation also suggests an alternative formulation, open quote, ensuring that the mission parameters related to the critical functions of laws cannot be modified by the system without context appropriate human control and judgment, end quote. Finally, Chair, I would like to appeal through you to request, to kindly request to all delegations that may suggest deletion of language in this section and others to justify the grounds for such proposals and if possible to offer alternative solutions to reflect the concerns expressed therein. Thank you.

Let me thank the distinguished representative of Brazil for sharing some proposals and indeed let me subscribe to your remarks that it indeed is helpful for, I think, for the room, but especially for the Chair and his team to understand why you would like to delete a certain part already knowing, of course, that other delegations in the room might be attached to it, so deletion is often no option. We have to find a solution in changing the text. Next on my list is Israel, to be followed by Belgium. But let me now pass the floor to the distinguished representative of Israel. You have the floor.

Thank you so much, Chair, and thank you once again for the work and for the facilitation of this complex yet important discussion. With your permission, Chair, we will do a quick reaction regarding comments we heard this morning on section two before moving on to section three. We apologize for the delay, but we were assimilating the valuable comments by other delegations, and unfortunately for the moment, no algorithm to do that for us. So, regarding section two, first with regard to paragraph four, we have heard suggestions from other delegations. In our view, the original text of this paragraph should be maintained given its status as a historically agreed language and the need to avoid implying that Martin's clause means something different than it has meant so far. Second, Article 36 itself states that in the development of a new weapon, a high contracting party is under the obligation to determine whether its employment will be prohibited by protocol or by any other applicable rule of international law. Meaning the article does not address legal obligations that instruct how to develop weapons, but rather addresses the development phase as a phase where future use is examined. This is an important distinction to maintain. Although Article 36 mentions development, it maintains IHL's nature as use-focused. Hence, we do not categorically reject any reference to development, but such reference should be confined to the context of legal reviews of weapons as does Article 36. Coming now to section three, looking at the third paragraph in section three of the rolling text, the paragraph in the current versions presents an imprecise citation of Article 51 of the Additional Protocol to the Geneva Conventions. We recommend preserving the original language to ensure clarity regarding the legal obligation. Regarding the fourth paragraph in section three on the rolling text, We point that anticipation or effect is not a primary obligation under the IDHL, but rather a second order requirement intended to fulfil certain primary obligations. In the current text, this paragraph treats it as a primary obligation that can disqualify weapons, while this does not reflect the law. Moreover, The use of this term could read as an expectation for a degree of certainty that does not exist in reality. IHL requires to make a reasonable effort to try and assess collateral damage based on information reasonably available prior to the attack, as opposed to predicting it with certainty. Unpredictable collateral damage does not, as such, render an attack as illegal. The current wording does not reflect these important nuances and may be read as prohibiting many existing weapons or uses of weapons. The core point on anticipation effects, on effects that does have a possible legal implication on the legality of weapons is sufficiently covered by the reference to the possibility of limiting the effects of the use of a weapon. In addition, predictability is addressed in other places in the rest of the text. For all these reasons, we suggest that the words "anticipated end" to be deleted. In addition, we think that the phrase "in all circumstances" is unsuitable. The end of the current paragraph explicitly and correctly refers to dependency on the circumstances of use. We also request replacing the word "control" with "limited" to align the text with the terminology in Article 51 of the first Additional Protocol. Finally, regarding the fifth paragraph in section three of the rolling text, as explained in previous sections, we object to the language of "context appropriate human control and judgment" and think that there is an alternative language that could be suitable, and achieve consensus such as appropriate human involvement. Another important issue again is that human machine interaction is not a primary IHL rule, so we disagree with putting it on the same list with established IHL prohibitions. However, we can see it reframed as a tool to implement IHL. For these reasons, we suggest changing the text of the fifth paragraph to the following text. Appropriate human involvement with regards to laws must be ensured where required to comply with IHL obligations under the circumstances. We will stop here for the moment and come back with comments from paragraph six and on later on. Thank you, Chair.

Let me thank the distinguished representative of Israel for sharing the views of his delegation with regard to the first five paragraphs of box number three and making some proposals. Next on my list is Belgium. You have the floor, sir.

Given this is the first time that Belgium takes the floor in this meeting, I would like to begin by expressing our gratitude to the Chair, his team and the Secretariat for the excellent preparatory work. Belgium is committed to contributing constructively to this important meeting. On box three, our first remark is that we support adding develop before use in paragraphs one, two and four. Like was also, it was also mentioned by Ireland and Brazil. Secondly, on para 6a, we would like to add actions before effects so that ensure that laws and their actions and effects are adequately predictable, reliable, traceable, explainable. And this because, of course, an effect is a consequence of an action. And while it might be challenging to predict, record, and explain effects, the must have in our view is that the actions of the law, of the laws are predicted, recorded, and explained. Thirdly, we support the rest of the box, of course, but also limiting the types of targets, duration, geographical scope. So we support that part. Fourth remark, we would like to add an element that was also sent by the Belgian delegation to the Chair beginning of February, namely introducing requirements for timely intervention such as reprogramming by human user after activation. So this is an element that we would like to see added in this under this para 6c as one of the small eyes. Another thing that we would like to add has to do with an explicit mention of the importance of recording, for example, by adding a small eye, an element introducing requirements for a secure data recording system. And lastly, this has to do with concern about real time machine learning and let's say limiting the self thought actions by the laws that result from real time machine learning and proposal from our sites to introduce this is limiting the ability of the weapon system to implement machine learning in its critical functions in real time. This is also a sentence that we sent to the Chair early August. Thank you.

Let me thank the distinguished representative of Belgium for his kind words addressed to the Chair and for his proposals made. It's almost one o'clock, so I think we will then resume at three o'clock. to hear the views of other delegations with regard to the rest of, with regard to box three and also box four. Let me wish you all a nice lunch break. We'll see you again at 3:00 PM. This meeting stands adjourned.