2nd Meeting - 1st Session Group of Governmental Experts on Lethal Autonomous Weapons Systems 2025
Meetings & Events
Date: 3 March 2025
Language: English
Transcript: https://transcripts.un.org/en/asset/k1a/k1afmtiaoj
Transcripts available through this tool are created by using automatic speech recognition and are not official records nor official documents of the United Nations. Official records and official documents are available on the Official Document System of the United Nations.

---

Netherlands (Kingdom of the) · Chair · Robert Indenbosch [0:03]:

Excellencies, distinguished delegates, I would like to continue our discussion with regard to the rolling text, and more in particular with regard to box 1. We still have a number of contracting parties on the list regarding their remarks on the text. I would like to start with Sierra Leone to be followed by Germany, Peru, Israel, South Africa, and at any moment we have also Colombia, but the person who has to deliver the speech is not yet available, so we start with Sierra Leone. May I pass the floor to you?


Sierra Leone · Joseph M. Conte [1:08]:

Good afternoon to the following session. I'm Joseph M. Conte from Sierra Leone. Thank you, Mr. Chair. In lieu of making general statements, my delegation would like to make the following general commentary in respect of box one. of the ruling text. Sierra Leone welcomes the ruling text as a significant achievement in our collective efforts to address the humanitarian, legal and ethical challenges posed by autonomous weapons systems. We recognise the substantial work undertaken to find common ground and appreciate that The text reflects many areas of provisional consensus. The current exercise represents an important step in developing elements that could serve as a foundation for future negotiations. As such, Sierra Leone emphasizes that we must not prematurely weaken substantive protections at this stage. What we preserve now will shape the starting point for eventual negotiations on a legally binding instrument. Mr. Chair, Sierra Leone believes the rolling text is moving in a positive direction by expressing a set of draft rules on which there is broad agreement. The suicide nature of the text is advantageous as regulating autonomous weapon systems will require clear core rules from which more detailed operational context can be developed. While acknowledging that future refinement is needed, We urge all delegations to recognize the positive aspects of the CHEERS text and build upon this foundation rather than diminish or narrow it. The element need not be perfect to serve as a basis for future negotiations. As a co-sponsor of draft protocol six, She really notes the significant convergence between that protocol and the rolling text, particularly regarding the application of international humanitarian law, the need for human control, prohibitions on certain systems, weapons reviews, human responsibility and risk mitigation. While the rolling text remains open to different normative approaches, we believe it could benefit from a more explicit reference to the possibility of negotiating a binding instrument as draft protocol seeks. Now coming to the area of characterization of autonomous weapons system, Sierra Leone supports refining it as follows. We reiterate our recommendation to delete the term lethal throughout the text as it has no basis in international humanitarian law. International humanitarian law protects civilians against death and injury and also protects objects. If retention of lethal is necessary for consensus, then we support including an additional paragraph clarifying that the term encompasses, as proposed by the ICRC, the ability to cause injury to or death of one or more persons and or damage to, destruction of, or neutralization of one or more objects. Regarding the cumulative application of identify, we share the points raised by Ireland, Brazil, Mexico, and Switzerland, and we emphasize the importance of maintaining the and between identify and select in order to prevent exclusive systems from the characterization where human users only perform identification while autonomous selection and engagement of targets occurs. In conclusion, Mr. Chairperson, We believe the reference to integrated combination should be revised to avoid implying all functions must be contained in a single physical unit, as autonomous systems may operate across distributed components. While we undertake the text intends to capture systems whose components may not be physically integrated, but collectively perform autonomous functions, the current wording could be interpreted otherwise. We have concerns that this terminology might inadvertently exclude distributed systems where components work together without physical integration. In this regard, we support ICRC's proposal to replace the term integrated combination with a phrasing, a system including one or more weapons. At this point, I want to say thank you for listening and thank you. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [8:04]:

Let me thank the distinguished representative for his comments and suggestions. Next on my list is Colombia. Let me pass the floor to the distinguished representative of Colombia, to be followed by Germany. But first Colombia, you have the floor.


Colombia [8:21]:

Thank you very.

Much, Chairman, and I do apologise for having forced a change in the order of the list of speakers. To start with, I would like to express our support for your role as chairman of this group. Furthermore, we would like to extend our thanks to those countries in the G16 who have expressed their opinions in the course of this morning's meeting. We have all been looking to make progress on a legally binding instrument on this particular topic. As far as this first box is concerned, Colombia has two brief comments she would like to make. First of all, for us, this issue of human control over this type of weapon is extremely important. The matter is in our own internal legislation, and that is why we support the inclusion of a reference to human control. We think it is essential that we have safeguards which ensure that there is human presence in the taking of important decisions. particularly on the use of force. We want to have a traceable and solid chain of command and responsibility where human beings do have an important role. Our second comment comes as a result of what we have heard from other delegations in the room, and it is about whether or not we include the word lethal'. We can support the proposal which was made firstly by Brazil, I think, and then backed up by other States, that we have an explanation as to exactly what it is this term is referring to. Somebody, I said, I think, or several delegations this morning before lunch said that if we're going to include this word, then we do have to have a definition of it, and that is what we would also support. We think the text of the ICRC would be appropriate for use here. because that specifies exactly what this concept actually means. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [10:35]:

Let me thank the distinguished representative of Colombia for her remarks and views and suggestions. Next on my list is Germany, to be followed by Peru, Israel, South Africa, the ICSC and Italy. But first Germany, to be followed by Peru. You have the floor.


Germany [10:55]:

Thank you, Mr. Chair. Mr. Chair, the German delegation would like to express its sincere gratitude to you and your team for the profound and intense preparations in the run up to this week and for the highly valuable guidance that you provided during the intersessional period. The constructive spirit in which our informal discussions in January were held confirmed our view that the GGE is the only appropriate forum for our discussions on lethal autonomous weapon systems. The German delegation will try to show as much flexibility as possible and work actively towards fulfilment of the GGE's mandate. We are of the view that the current version of the rolling text captures the essence of the intense discussions that this group has had in the past years. It strikes a balance between the required level of ambition, taking into account the relevancy and urgency of finding consensus on a prohibition, respectively regulations, of lethal autonomous weapon systems, on the one hand, and the difficulties associated with the complex nature of questions related to law. It also demonstrates how much substantial progress this group has already made and how much convergence we built up in recent years. This makes us confident we do see a chance that the GGE can fulfil its mandate in time. As for part one, Germany in general supports the suggested working description of a lethal autonomous weapon system. Given the enormous difficulties that are related to the question of a definition, we highly commend the suggested wording. Nevertheless, we suggest inserting a further' at the end of the last sentence so that it reads without further intervention by a human user in the execution of these tasks'. In our view, a LARS is a weapon system with autonomous functions that after its activation does not allow further intervention by a human user to exert further control over that system. A LARS, according to our definition, is a weapon system with autonomous functions that is completely removed from human control and operates outside a responsible chain of command. Without further, the characterization would apply only to fully autonomous weapon systems, rendering the two-stage approach meaningless. We are looking into the proposal to include the definition of lethality by the ICRC. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [13:07]:

Let me thank the distinguished representative of Germany for her remarks and now pass the floor to Peru, to be followed by Israel. But let me first pass the floor to the distinguished representative of Peru. You have the floor.


Peru [13:23]:

Thank you very much, Chairman. I would also like to start by thanking you and the Friends of the Chair for the considerable progress which you have managed to achieve with the circulation of this version of the rolling text. My country is of the opinion that this text reflects the majority of the concerns which were expressed last year and furthermore is looking at areas where there are still divergences as well. Consequently, Peru, as a member of the Group of 16, as has been mentioned by quite a few other speakers who preceded me on the floor and who are also members of the G16, believes that we need to start negotiating process on a legally binding instrument. As to these first two paragraphs in box number one, my delegation would like to make its position clear in respect of the question of lethality. The position of the Group of 16 is clear and is reflected in the working document which we submitted in 2023. We're flexible. If the word lethal and the concept is retained here, then we want it to be clearly defined. And with that in mind, we feel that the proposal to use the ICRC language is a good step forward. In addition, We've heard with interest the proposal for a paragraph 1B bis, which I think was made by the delegation of the United States. We've been having a look at that and thinking about it. And the other point I wanted to mention is the one referring to the last part of paragraph one where it says, and I read in English, without intervention by a human user. Our preference, as has also been made clear by other members of the Group of 16, is to maintain the idea of significant human control. Interpreter apologizes, meaningful human control. Thank you. Thank you, Chairman.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [16:20]:

I thank the distinguished representative of Peru for his intervention and for his suggestions. Next on my list is Israel, to be followed by South Africa and the ICRC. Let me first pass the floor to the distinguished representative of Israel. You have the floor.


Israel [16:39]:

Thank you so much, Chair. Allow me first to express our great appreciation to your work and leadership, including to thank for their recent informal meetings which we found very helpful in focusing the discussion on the key issues of disagreement and bringing us closer to convergence. Concerning the text in section 1, without expressing a definitive or comprehensive view of the matter of the characterization, we have several comments. First, as conveyed by several delegations this morning, such as India and the United States and others, it is Israel's position that any characterization of laws must focus on lethal autonomous weapon systems and not just on autonomous weapon systems, in accordance with our current mandate. The inclusion of no lethal force could lead to broadening the debate to encompass fields which fall outside the purview of the CCW mandate and expertise, such as law enforcement contexts. In addition, dropping the lethality element, as some suggest, may also divert us from the most significant issue at hand, which already presents substantial issues, that we have still not resolved, it will only complicate even further our attempts to reach agreements. Therefore, we believe that lethality element must remain. As for the suggestion to include a definition of lethality in the text, Chair, previous meetings showed us that the meaning of this term is subject to significant disagreements. in that the attempt to define lethality is complex and not yet fully developed. We are concerned that this issue could draw this forum away from consensus, so practically thinking, we should remain cautious in pursuing this attempt. Now, regarding the first paragraph section in the text, to maintain clarity and focus within our mandate, the discussion should keep centering on weapon systems. an issue which has been complicated enough and required concentrated attention rather than making technological enabled components a distinct subject to focus on. Therefore, we prefer retaining the older and simpler articulation of weapon system without the additions suggested in the previous meeting. An alternative for that will be to make the currently suggested text clearer in its focus on weapon systems by changing the word integrated to integral and replacing the words enable the system with constitute together a weapon system. In addition, like previous other delegations such as Singapore, the Republic of Korea and others, we also request clarifying that the functions identify, select and engage should be cumulative rather than alternative. This will express the uniqueness of laws that were the focus of the CCW's discussion so far. Sufficing with identification as a single requirement or with selection as a single requirement will not even qualify as a weapon system. But even identification and engagement cumulatively wouldn't be enough. For example, Such a formula could cover common missiles with homing guidance that are launched by humans to the direction of a specific target and lonely pinpoint the target in their journey based on the identification of certain signal. These are munitions that exist for decades and raise no real issue of autonomy. They are not the center of the forum's attention and interest. For this reason, we think that the word "or" should be omitted and the word "then" be added before "engage" to make it clear that the engagement is also cumulative. Finally, any inclusion of the function of selection and identification in the definition of laws should refer to the independent exercise of identification and selection of targets an application of force in order to discern laws from automated weapons that are not in the focus of our discussion, such as precision-guided munitions. We therefore suggest adding the word independently before identity. The text will then read that once activated can independently identify and select and then engage a target and so on. Regarding the second paragraph in section one of the rolling text, we suggest starting the paragraph with a disclaimer similar to the one found in the political declaration of responsible military use of artificial intelligence and autonomy. We've lived equal help to get us to consensus due the diverging views regarding the definitions and characterizations. With this disclaimer, the text will read acknowledging that concepts of autonomy are subject to a range of interpretations and are a developing issue, the above description and so on. We also think that modifications should be changed to alterations to express more flexibility. We will stop now. I award that the intervention was quite long. So if any delegation will want to have it in written, we will be happy to provide it. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [22:49]:

I would like to thank the distinguished representative of Israel for his intervention and suggestions, and the Chair indeed appreciate also that we received your proposals already in writing. And if other delegations would like to have that, please make contact with the Israeli delegation. We have on our list South Africa, ICRC, Italy, Cote d'Ivoire, Bulgaria, but first South Africa. You have the floor.


South Africa [23:22]:

Thank you very much, Mr. Chair, for giving me the floor. South Africa wishes to congratulate and appreciate the manner in which you have steered the work of the GGE since the commencement of the renewed mandate, and we hope that the GGE will advance our work and yield substantive outcomes leading to the review conference in 2026. South Africa underscores the importance of a clear and precise characterization of lethal autonomous weapon systems, While the current text references impact, we believe it is essential to ensure that any definition does not lack a comprehensive analysis of lethality. In this regard, we see merit in the proposal made by Brazil for an additional paragraph that provides greater clarity on this aspect. Furthermore, South Africa stresses the necessity for meticulous consideration regarding human involvement in laws. If the approach entails ensuring a human is in the loop at all times, this requirement must be explicitly stated in the text. Conversely, if such a stipulation is not clearly defined, the characterization should acknowledge the existence of fully autonomous lethal weapon systems where human intervention is non-existent, while recognizing that this could broaden the current discussion. From our perspective, any characterization of laws must affirm the critical decisions, particularly in identifying, selecting, and engaging targets must always remain within human control. The operator or user must retain significant decision-making power to ensure that the responsibility and accountability is an obligation that cannot be left to the machines. This principle is fundamental to maintaining compliance with international law and ethical standards. South Africa looks forward to further discussions on refining these elements to ensure a balanced and comprehensive approach. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [25:29]:

Let me thank the distinguished representative of South Africa for her intervention. Next on my list is the ICSC, to be followed by Italy, but first the ICSC. You have the floor.


ICRC [25:43]:

Thank you, Chair. And from the outset, the ICRC would also like to express our appreciation to you, Chair, to your team, and to the friends of the Chair for the hard work and consultations on the rolling text. The text reflects many constructive contributions made by high contracting parties and observers, and it contains the structure and substance of a set of elements that we see as a good basis for negotiation of a legal instrument to regulate autonomous weapon systems. In this regard, In line with the clear calls made by our president and the Secretary General of the UN, as well as a number of states, the ICRC remains convinced of the need for urgent action on autonomous weapon systems, and we urge high contracting parties to capitalize on the progress that is reflected in the rolling text and to conclude the group's work this year with a strong set of elements that can form the basis of negotiations for a legally binding instrument by 2026. We also echo the comments made by the distinguished representative from Pakistan, noting that the group is not currently engaged in formal negotiations, but rather is compiling elements. And so a reductive approach of deleting text should be avoided. In the context then of the working characterization, we see it as imperative that the group retains a broad characterization of autonomous weapon systems that captures all systems of concern. Of course, this does not mean that all such autonomous weapon systems will be prohibited. Indeed, as was outlined eloquently this morning by the distinguished delegate from Switzerland, that is the function of the two-tier approach that follows in box three. Only certain types of autonomous weapon systems are to be prohibited, with the remainder subject to restrictions. We echo the comments this morning from Pakistan, Finland, Costa Rica, the Dominican Republic, Sierra Leone, and the Philippines, among others, reiterating our recommendation to delete the qualifier of lethal throughout the text, as IHL protects civilians against death and injury and also protects objects. That said, in the spirit of working towards consensus, we have proposed incorporating a common understanding of the meaning of the term as a standalone paragraph, and the text of this was already set out this morning in the intervention by Brazil. We noted with interest the proposal of the United States in this regard, and we do consider an additional one bis a logical solution. We would be concerned to ensure that the clarification captures weapons that may not be capable of causing death, but they are capable of causing unnecessary suffering and superfluous injury, or of damaging, not only destroying, objects. But we do think the structure of a one bis seems a sensible proposal. On the point as to the cumulative inclusion of identify, we share the concerns highlighted by many delegations already about unduly narrowing the scope, and we emphasise the importance of retaining or before identify so this function is not made fully cumulative. Like Brazil and Austria, we would be interested to hear more from delegations supporting a cumulative approach as to their understanding of the term identify and whether they have in mind any particular systems that should be excluded. And we note the useful comments already from the United States, Singapore, and Israel in this regard. For the ICRC, the term identify would refer to the process of determining a general target profile or a list of potential targets to be input to a system. In contrast, select involves matching a specific person or object against that pre-programmed target list or profile. That is selecting someone or something for a strike. For the ICRC, a weapon system should fall within the scope of an instrument even if the identification of a type of target is carried out by humans. Because the key point is to capture all systems where a human does not select the specific target and doesn't then engage that target with force. We read the current and/or formulation as making the two functions of select and engage mandatory. We do not read it as making select interchangeable with identify. Finally, we support retaining the term human user as outlined by the distinguished representative from Australia. This ensures that the characterization covers not only operators, but also all commanders and others responsible for planning, deciding upon, and executing attacks as required under IHL. I thank you, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [30:46]:

Let me thank the distinguished representative of the International Committee of the Red Cross for sharing her comments and suggestions. Next on my list is Italy, to be followed by Cote d'Ivoire and Bulgaria. But first, the distinguished representative of Italy, you have the floor.


Italy [31:08]:

Thank you, Mr. Chair. At the outset, along with the other delegations, allow me to express our appreciation for the tireless work you have conducted over these months. A range of positions and views are present in this conference room, and through your efforts, they've been nailed down to the current rolling text, where we see several areas of possible convergence. We had the opportunity to highlight in the past that when we were tasked to further consider and formulate by consensus a set of elements of an instrument, we looked at the future instrument that should try to adopt a functional approach, focus more on a scope of application rather than physical characteristics. Looking at this definition, we would like to present some observations. In our view, a lethal autonomous weapon system is a system that, once activated, can identify, select, and engage targets without further human intervention. The definition's principal characteristic is the role of the operator with regard to target selection and engagement decision, rather than the technological sophistication of the weapon system. For this reason, we would advocate for a definition of lethal autonomous weapon systems that emphasize decision-making autonomy in critical functions, and we would express preference for identify, select, and engage with the addition of once activated and without further human intervention. Regarding the notion of lethality, for us, the mandate of this group is clear and it pertains to lethal autonomous weapon systems. Having said that, we are open to consider a point one bis providing further explanation on this content, as highlighted by the distinguished US delegation this morning. And we think that that wording provided could be a good starting point for further discussion. Finally, we also favor your proposal to not list the types of systems that do not follow within the scope of definition, namely and not limited to systems that operate in an automatic mode without human intervention after activation as a certain self-defense systems. Such list would in fact hardly be exhaustive and future proof for further developments. And I thank you, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [33:49]:

Let me thank the distinguished representative of Italy for her comments and remarks and suggestions. Next on my list is the Cote d'Ivoire to be followed by Bulgaria. But let me first pass the floor to the distinguished representative of Cote d'Ivoire. You have the floor, sir.


Côte d’Ivoire [34:08]:

Thank you, Chairman. Good day, I would like to thank you for the way you are steering our work and we would also like to pay tribute to the efforts which have been made by all delegations involved in this discussion. We would also like to convey our gratitude to the Friends of the Chair for their active contribution to this process, which has helped us to make headway on what is a critical issue. As to box 1 on characterisation, in substance my delegation has taken note of the opinions expressed on the notion of lethality, particularly in the characterisation of laws. More than one party has expressed some doubts about using the word lethal because some weapon systems can cause irreversible harm without actually killing somebody. and the mandate of the GGE is not to exclude lethality. Like several other States, we support the solution consisting of maintaining the reference to lethality while taking a nuanced approach, perhaps based on the ICRC definition. On paragraph That says that the characterisation could change and modifications will have to be made in the future. We would like to see this maintained because when technological developments are taking place so fast, it is a good idea to be careful about what might happen in the future. Generally speaking, we support the proposal to add headings to the sections or boxes or chapters, whatever you call them, as the US suggested, so as to make the text more reasonable and comprehensible. We will also contribute fully to this work in accordance with the mandate of the group, and we would like to welcome the progress made with this rolling text. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [36:00]:

Let me thank the distinguished representative of Côte d'Ivoire. Next on my list is Bulgaria, to be followed by Austria. But first, let me pass the floor to the distinguished representative of Bulgaria. So you have the floor.


Bulgaria [36:14]:

Thank you, Mr. Chairman. Mr. Chairman, this is the first time our delegation takes the floor. Be assured of our full support for you and your team's endeavors as you assume this demanding role for another cycle in 2025. Bulgaria welcomes the convening in January of three rounds of informal consultations with the aim of facilitating constructive.

Discussions on the latest version of the rolling text.

We believe the intersessional period was used in a productive way for coordinating among delegations to research convergences and differences. Bulgaria highly appreciates the Chair's approach to streamline the GGE's work by focusing on specific texts contained in the rolling text on defining and formulating elements of an instrument. With regard to section one and characterization, we have some short remarks. We believe that the current text reflects rather well the intense discussions the group held in 2024. And on para one, we support the inclusion of further before intervention, as pointed out by several delegations already. We support the general formulation of para two as well. And regarding the notion of lethality, we understand its controversial nature within the GGE. And to clarify, Bulgaria doesn't consider lethality to be an inherent characteristic of an autonomous weapon system. Rather, it is our understanding that lethality is in effect. Still, we can be flexible on this issue.

Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [38:00]:

I would like to thank the distinguished representative of Bulgaria also for his flexibility with regard to the text, um, next on my list is Austria to be followed by Ukraine and Brazil. But let me first pass the floor to the distinguished representative of Austria. It's the second time you take the floor. No problem at all. I welcome that because this should be an interactive discussion. And I also remember that a number of delegates already indicated that their intervention was a first intervention with a limited number of points and they had more to share with us. So I really look forward to hearing the rest because I need to have a full picture of your views in order to be to, to, um, to take them into consideration, uh, when making a next effort with regard to the, uh, rolling text, but let me now pass the floor to the distinguished representative of Austria, you have the floor.


Austria [38:58]:

In good diplomatic practice, I still want to apologise for taking the floor a second time and share our views also in light of the discussion. They come down to, I think, the two main points that we have been discussing this morning about lethality and about what is called now the cumulative approach on the functions in the working characterisation. I have followed the discussion a little bit with mixed feelings because, on the one hand, I feel that we are repeating a discussion that we already had and I also have to state that since the previous years, many delegations seem to have switched their position, especially on the lethal discussion, which is a big surprise. Trying to understand the switches in positions, I think, and this has already been hinted by the ICRC, that it's probably touching upon certain weapon systems that might be touched by this working characterization as it stands right now, and probably the same is true for identify and select and/or engage. What we need now, and this is why I have mixed feelings, is explanations. We need to know what exactly is the problem, why has lethal so far only been an issue for five or six delegations and now it's a larger number, what kind of weapon systems to create this kind of concerns with the working characterization, and how can we resolve it in this text, because it is a fact that we have different views on this. Today is an opportunity to resolve this and I welcome any kind of explanations that our care can provide. Something and maybe to explain our understanding of the working characterization is that it should be as broad as possible. I understand that there is some indication, some intention to limit it already now and to avoid discussion about certain weapon systems that already exist or that seem too very dear to some states. But our understanding on how negotiations on such a text should work is that first we have a working characterization on what we are talking about, and then only in a second stage we go into excluding certain elements. And this is also how many of the protocols of the CCW have worked. First, we set out the working characterization, Then during regulation and prohibition phase, we would look into what could be okay under the current circumstances. And then only in finalizing a possible instrument, we look into possible exclusion. And this is, I think, a best practice and also how we think we should solve it. What is also important when it comes to the lethal discussion is that we should not We should really take into account also that there are weapon systems that are designed not to kill people, that they would not fall under our work and would be sorry, I'm getting ahead of myself, I have some strange notes here. The problem is that in our understanding there are also weapon systems that should fall under a possible instrument and under our work as it is tradition the CCW that would also maim and injure people. and that we should not exclude this kind of even violation of IHL, because this would mean that in the end we would have such weapon systems that are not lethal but that could still work in a fully autonomous way. So we are creating different categories. We are creating those that are designed to be lethal and we are creating those who are not designed to be lethal. And this creates a certain risk that we have different legal systems for those two kind of categories, which in our understanding both fall under IHL and under the scope and work of the CCW. So this is just another concern that we are having here. We would also be interested in more information about the select and/or engage issue. In our recording, in my memory, we had this discussion already last year and this was one of the solutions that we go to identify and/or select because there was a misunderstanding or different perceptions on identify and select, what it means, is it the same, do they have different qualities. The solution was that we should keep both and put an and/or because there is a certain overlap. The original way of dealing with this and the one that we also supported from the very beginning was to not have identify and just talk about select and engage as it was the case for the ICRC definition for a very long time and also in the first drafts that we had on the screen. So this could be a possible solution if this is not resolvable at the moment right now. As mentioned at the beginning, we will be interested what exactly is the problem in the different functions and what kind of weapon systems you are talking about that would create those concerns, because in the end, as mentioned before, we would not like to see systems that could be separated and then not fall under this scope. In the end, the overall aim should be to be as broad as possible. And we would be also interested what kind of system would identify and engage or select and engage and there at the same time would not be an autonomous weapon systems or lethal autonomous weapon systems. So this kind of explanation is necessary and we definitely need to find a solution for that. I think those are my main points right now. I had some others, but they are somewhere in my notes, so apologies and thank you for listening to me.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [44:44]:

Let me thank the distinguished representative of Austria for sharing his views and also reminding us of previous discussions. Let me now look at the list. I have three more requests for the floor. I have Ukraine, Brazil and Greece, but let me now first pass the floor then to Ukraine. Sir, you have the floor.


Ukraine · Technical expert [45:05]:

Mr. Chairman, As a technical expert, I confirmed that our delegation agrees with this definition and we see that this definition should be included little because it is our mandates and we have task considered only little autonomous system. Our delegation support position of United Kingdom regarding this definition. And in my opinion, this definition can be take as a base for future improvement. And in future, we can be return again to improve this definition again. And my proposals now is regarding position of to function, identify and select. In my opinion, select should be first because after select, it can be identify it is friendly or enemy targets. And second, uh, comment regarding position of USA about replace human user to operator, I think that it is more better because in future as operator can be considered not only human, uh, but also artificial intelligence. And in this case, it will be operator more expanded definition. And the last comment regarding using more weapons, it's in my opinion very important because swarm technology in future will be implemented in combat field. And in this case, we should be considered not only one weapon system, but also integrated swarm. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [47:00]:

Let me thank the distinguished representative of Ukraine for sharing his views and proposing suggestions for the text. Next on my list is Brazil, to be followed by Greece and the US, but first Brazil. So you have the floor.


Brazil [47:17]:

Thank you, Mr. Chair. I also wanted an opportunity to intervene again on this issue to react to some of the interesting ideas that we have heard this morning and early in the afternoon. Regarding the U.S. delegation's suggestion on the use of operator instead of user, my delegation would be flexible with that, with those options. Our preference, however, would be to retain the word human before either because exactly of the problem alluded to by my previous -- the previous speaker, which is the possibility that this in the future operator or user could be understood to be something other than human. Regarding the issue of integrated combination, my delegation had understood this as not necessarily a physical integration, but a functional integration, and this continues to be our understanding since autonomous systems can be composed of separate components that are not necessarily contained in one single container, so to speak, but nevertheless function as a system. I allude, for example, to the cases of close-in weapon systems that are not self-contained but are integrated with the radar systems of the vessel in which they are installed but nevertheless work very much like an autonomous or can work as an autonomous weapon system. Regarding the points made by Russia and by the French delegation on the issue of temporal dimension, understand that the temporal dimension which autonomy kicks in, so to speak, I think that the solution that you have proposed is, as we said in the morning, very elegant and adequate, but we could live with the suppression of in the execution of these tasks if it's deemed to be not helpful. But as we said, we prefer the formulation that you have originally put forward. On the issue of lethality, this morning my delegation stated that if the term is to be retained, it needs to be defined, and this for a very simple reason, which is the fact that we have different understandings in this room about what lethality means. And our debate since the beginning of the session has demonstrated this amply. Some delegations seem to understand lethal force as destructive or kinetic force. Some espouse a very restrictive understanding that would basically restrict the scope of this characterization to anti-personal weapons. On this point, I would like to recall the very precise intervention made by the ICRC, which highlights the fact that IHL does not only protect persons, but also protects objects. So I think that this highlights the need for clarity. If we are to retain this term, it needs to be clearly understood and cannot be restricted only to anti-personnel functions. In this regard, I would like to point to the very interesting intervention made by my distinguished Swiss colleague this morning, suggesting that concerns that underpin these proposals from some delegations about treating anti-personnel and anti-material weapons in the same way could be dealt with in the further sections of this text. And regarding the cumulateness of functions, also something that we have discussed at length. Again, my delegation sees no reason to include identify as one critical function and we see that there is a risk in doing so. given the fact that even if a human very perfunctorily acts on the identification phase, but the selection and engagement are left to a weapon, this would be the case of almost all loitering weapons in existence, these would be excluded from our discussions if we are to adopt a cumulative approach. So in this regard, I really would appeal, like my Austrian colleague, to delegations to clarify what they mean by identify and to be clear about which systems are of concern and that they wish to exclude. And on this, again, as Switzerland said, it is important to recall that this characterization is not about identifying the systems that are to be prohibited, But the systems that are of concern to our discussions, prohibitions will be dealt with in section three and four. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [52:41]:

Let me thank the distinguished representative of Brazil for further developing his views and reacting to the comments of other delegations. I have now on my list Greece to be followed by the USA. But let me first pass the floor to the distinguished representative of Greece. You have the floor.


Greece [53:03]:

Mr. President, at the outset, I would like to congratulate you on your efforts and leadership in guiding these discussions toward a tangible outcome and assure you of our delegation's full support in your endeavors. In our view, lethality is a core element, and as other delegations have outlined, it should be reflected in the working definition. Furthermore, we also support that all three prerequisites, identify, select and engage a target, must be met to define a lethal autonomous weapon system. The absence of a clear definition of what constitutes a weapon system affects the development of a common understanding of what defines a lethal autonomous one. From this point of view, we support the current wording of the need for one or more weapons and technological components, but as the US delegation outlined, we must define what the technological components are, such as sensors, software, et cetera. These elements, from our perspective, are essential for establishing an effective understanding of what constitutes a lethal autonomous weapon system. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [54:08]:

Let me thank the distinguished representative of Greece for his comments. Next on my list is the USA. Sir, you have the floor.


United States of America [54:19]:

Thank you, Chairman. So we also wanted to react to a number of points that were made in the room over the course of this morning and this afternoon. On lethality, I think we would agree with the diagnosis that was just presented by our colleague from Brazil on how it is that different delegations are approaching this differently. I think for us, the cure is still captured in the idea of having an additional an additional sentence along the lines that we had proposed. I think there seems to be interest in the room in exploring that, but there does seem to be a bit of debate now about the approach that's reflected in the ICRC proposal, which has been circulated in writing. People have had a chance to look at it and think about that. We did want to just draw out a few of the differences between what we proposed and what's in the ICRC proposal, just because for us, I think the ICRC proposal is a bit overbroad. I think if you focus on the language of the ICRC proposal, they have an expansive definition that is wide enough to encompass, among other things, an ability to neutralize an object. I'm boiling it down to that concept because that is one of the available readings of the ICRC definition. I think for us, the idea of neutralizing an object, if you divorce it from kinetic action, could encompass a lot of different types of different means and methods, including electronic methods or cyber methods. And so for us, I think this is a little too broad. Uh, I, I, we welcome the, um, the, the proposal in the sense of, of promoting a rigorous discussion of what it is that we need to do to create this further definition. But just to go back through our language, which I know we only read, so people haven't had a chance to focus on it, but we do begin with the idea that lethal force is force that is capable of resulting in death. But then the next sentence clarifies that the fact that the laws can be used in circumstances that do not result in death such as to destroy objects or to cause injury, does not exclude it from this characterization. So we do try to get at what we think are the essential impacts in the world and want to make clear that it is potentially broader in the type of impacts. But just to differentiate between our proposal and their proposal, I think we would be more comfortable with something that is focusing more on kinetic operations. Second, Chairman, on the debate about identification, selection, and engagement, as we said this morning, to us, the and/or seems to indicate that identification and selection are not essential characteristics, that they're interchangeable. And we don't see them as interchangeable. We think that they're both essential characteristics. And just to give an example to help illustrate the point, the capability of the system to identify a target is necessary to both target selection and target engagement functions. To select a target, the system needs a capability to perceive the target as a possible option for engagement. To engage a target, the system needs to be able to direct force against the target, which necessarily involves the ability to locate the target in the environment. Third, on the discussion that we're hearing surrounding the idea of a qualification for the type of human intervention that would result in the weapon being considered autonomous, I think we've heard significant or meaningful human intervention. I think we've heard both of those terms. For our delegation, this would add considerable uncertainty. It would appear to leave to the discretion of individual countries the decision of what, in their view, constitutes significant or what, in their view, constitutes meaningful. And from our perspective, this would make it unclear what systems are covered and what systems are not covered. And from our perspective, it would be inconsistent with the main objective of this particular section, which is to achieve clarity about the types of weapon systems that fall within the scope of our definition. Finally, we're very sorry to have opened the door on this vigorous debate we're having about headings or no headings. I guess what I would say on this point is that to the extent that some delegations are focused on, at this stage in the proceedings, on where these elements will ultimately land. I think the Russian Federation, for example, indicated that in their view, they see this as ultimately constituting what would amount to the conclusions and recommendations section of a report. And while that's certainly possible, I don't think we're at a point yet in our proceedings where we're in a position to make that final judgment. And so we just want to make sure that we're retaining the flexibility that we think is inherent in the mandate to work within this group to use these elements in a way that makes sense in the future. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:00:24]:

I would like to thank the distinguished representative of the US for further elaborating his views and proposals. Next on my list is the EU, to be followed by the Russian Federation. Let me now pass the floor to the distinguished representative of the EU. You have the floor.


EU [1:00:43]:

Thank you very much, Mr Chair. We thank you, Ambassador, for presenting the revised rolling text being the basis for substantial discussion that we are having. At the outset, I would like to reiterate that the European Union is fully committed to delivering and fulfilling the mandate of the group as soon as possible, preferably before the end of this year. We expect that a set of elements of an instrument, without prejudging its nature, and other possible measures to address autonomous weapon systems will be agreed in order to move forward. We stand ready to contribute to an ambitious and substantive outcome which reflects the urgency of the matter, starting with characterisation of lethal autonomous weapon systems. With regard to box 1 and characterisation presented, we are of the view that LWS should be characterised as systems that, once activated, are able to identify, select, track and apply force to targets without further human intervention. Thank you, Mr Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:01:57]:

Let me now pass the floor to the Russian Federation. Sir, you have the floor.


Russian Federation [1:02:03]:

Mr President, as a part of the interactive discussion which has arisen between the delegations and which, without a doubt, we welcome, we would like to offer two brief comments following the comments that were voiced by a number of delegates. First, with respect to the concept of lethality and the proposal from a number of delegations to use as a compromise the definition from documents of the ICRC. In our opinion, this attempt to use that ICRC definition in its very essence is the equivalent of taking the word lethal out of the definition, because it explains and reduces lethal autonomous weapon systems to simply autonomous weapon systems. We see no difference in the definition given by the International Committee of the Red Cross between the two concepts of lethal autonomous weapon systems and autonomous weapon systems. In our opinion, when we created this group back in 2016, we were specifically talking about the topic to be discussed here to be weapon systems built on the basis of new technologies in the area of lethal autonomous weapons systems. And I would underscore lethal autonomous weapon systems. Now, if we wanted to discuss autonomous weapon systems, there's no reason why we could not agree to do so. Just as across all of those years there was no barrier to us agreeing to discuss autonomous weapons systems when we were coordinating the mandate of this very group, and that includes the mandate which was agreed in 2023. However, what we agreed upon was a mandate which did not cover autonomous weapons systems, but lethal autonomous weapons systems. placing that category into a special category requiring discussion within the framework of the GGE. And now, in the middle of the discussion on this topic, the attempts by a number of delegations to move the discussion into a different discussion about autonomous weapon systems and to move away from discussing lethal autonomous weapon systems, those attempts appear to us to be extremely counterproductive. We very much hope that this is not an attempt to overhaul the very mandate of the intergovernmental group of experts, and to move the discussions away from the issues at the heart of the agenda to instead discuss some other issue. Our delegation continues to be guided strictly by the existing mandate, and we assume that we are here to discuss lethal autonomous weapon systems and not autonomous weapon systems in general. the very concept of lethality, as it is in Russian, implies causing death to troops, humans. There cannot be any other reverse translation. Lethality, as a very minimum, requires ending lives. We also understand that this term includes damaging health of humans. But if we're talking about causing damage to facilities, objects, then in our opinion that goes beyond the scope of the concept of lethality. And that would take our discussion back into the realm of discussing autonomous weapon systems in general. Because it's hard for us to understand any other autonomous weapons systems other than those who damage the health of persons, kill individuals or cause damage to structures and facilities. And this is why we consider that we need to talk specifically about killing and causing damage to human health, these two concepts are fully justified and can be reflected in the working definition itself. In addition, we would also like to touch on the topic raised by a number of delegations referring to the concept of meaningful human control. on many occasions over the past decade, almost a decade, have spoken about this concept. We've presented our national position, which remains unchanged as it has done throughout this period of time. We see no need to refer to this concept. It is not a legal concept. However, it does create complexities and challenges in defining what does meaningful mean and what does less than meaningful human control mean. In this area, We are guided by the principles and concepts coordinated by this group and clearly defined that human control must be provided for these weapons systems. For us, this is an absolute rule. and we do not think it is feasible or sensible to speculate about whether human control is meaningful or not. It seems to us that would once again draw us away from discussing the actual key problems around human control that this GGE faces and should discuss. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:10:37]:

Let me thank the distinguished representative of the Russian Federation with regard to further explaining his position, especially on lethality. It's clear that that discussion is not yet over. And you referred to meanings of lethality in Russian. I did a quick search on the internet on English. Yes, you get you get causing death or serious injury, but in many cases you also get causing serious damage. So I don't think it's as clear as it is, at least in English. I'm only a master of a few words in Russian, so I'm clearly not mixing into that discussion. But thanks for reminding you of your clear views and also views in the past. We have heard quite a lot of opinions on this issue. I am just looking around to see whether there are still delegations left that would like to have a first intervention or who would like to react to the views that have been given so far by those who have taken the floor. If not, then I would like to give a kind of a wrap up of our discussion, which is almost next to impossible, but I will try, because it is not more than indicating the further work the team of the Chair will do. But before I embark on that, I see a further request for the floor from Austria and Pakistan. But let me now pass the floor to the distinguished representative of Austria. You have the floor.


Austria [1:12:28]:

an attempt to continue this discussion as it is a key discussion and one that will not go away quite soon. I think what has been also mentioned by the Russian Federation is what is the heart of our agenda and what we are doing here. The issue of lethality has been with us for a long time and it has always been discussed. I think also at the very beginning in 2016, there was no full consensus on what it means and what it does to our mandate. This issue has never left us. It is not that we have agreed that this is one of the main issues and one of the main characterisations on our work here. At least in our understanding, it definitely does not exclude certain weapon systems. At the heart of our work in the CCW, to say this once again, is of course upholding IHL. And some of the issues that we will discuss and we have discussed last year are also accountability and responsibility. And it is hard to understand why a certain category of weapon systems that is lethal, that has the capability to injure, maim or kill people, has a different kind of discussion or at least no discussion when we talk about those that now seem to be excluded from our discussions than the one who can. So we are creating now, and here I have to voice a different opinion, we are creating now a new divide in our discussions that has not been there before and we have to state that this is not our interest. There cannot be an artificial differentiation between lethal and non-lethal weapon, I also have a question mark even on the definition on what a non-lethal weapon would be, especially in our context. In general, the CCW in most of its protocols also talks about non-lethal weapons. I mean, we are a convention that has a protocol on non-blinding laser weapons. So there is, by definition, something that is also talking about, not about lethality, but about other issues. And here I contradict myself about what I have said before. The other point that I wanted to make is that lethality, as it has been pointed out by the ICRC and also by others, has no basis in IHL. So if we now create a definition and use it potentially in a legally binding instrument, we are actually creating this basis, and there is a certain risk in that. It could have repercussions also in other discussions, even outside of IHL discussions, as has been pointed out by Pakistan this morning, and we should be fully aware of that. So we can have this one piece as part of working characterization, but when it comes to definition as part of a legally binding instrument, we have to be very careful and we will probably have a reservation on this until the very end. And this is also the point that we made this morning. This is something that relates to the rest of the text, it relates to our discussion. It is a working definition, it is not a final definition, and this is also how we should approach it. So for the moment, we could be fine with this kind of working solution, but there is certainly a reservation on including it in the long run. I thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:15:57]:

No, I thank you for further explaining your position. I have Pakistan. Let me pass the floor to the distinguished representative of Pakistan. Sir, you have the floor.


Pakistan [1:16:09]:

Thank you, Chair. Thank you for the opportunity to come again into the conversation and to react briefly on the conversation that has taken place. On the issue of lethality, in addition to what we said in the morning, we do see the problem as identified by our delegation of Austria just right now that although while staying true to the mandate of the group and using laws, there is a strong merit in having an explanation that reaffirms that what is covered in IHL and what is covered in IHL has been described by the ICRC very clearly that affects certainly in those cases which do not produce lethality are also governed under international humanitarian law. So in that context, I think the proposal that the delegation of the US presented earlier this morning was certainly interesting, and I think this is something that we could have a look at if the language is produced. In front of us, we do note that when we presented our working paper last year on the elements of an instrument on laws, we also had something similar suggestion in terms of what could be included in the characterization that regardless of the actual consequences of the use, whether the applied force results in lethal effects or not, it will be regulated under this protocol. And I think that is what the broader understanding of the room is. So we can have a look on that. We do understand that we made a similar attempt last year in the footnote, which became too weedy, so we would need to approach it in a very cautious manner. But this is going to be, I guess, one of the sticking points till the very end. We believe the definition should not be weedy. And what you have in the text right now, we should approach it very carefully in order not to further introduce any new elements in the text. We have heard some of the proposals in the morning conversation and we are reluctant in making some drastic changes in the text. On the issue of cumulative functions, for instance, We believe that the divide continues to exist whether we should have and or or between identify, select or engage. And I think you have chosen the right approach as of now. You have kept those two qualifiers there. We do understand that some delegates are perhaps wary that some systems would be captured which otherwise should not be captured in this definition, but for that purpose, the paragraph two, I, I think, uh, it, it serves that purpose that if we, uh, have to exclude in the future certain systems, we can certainly, uh, list them there, uh, in the future as well. And that brings me to the point that, that modification, the characterization could evolve in the future, not only just to exclude certain systems, but also to include certain systems which may not be present right now. So as far as we are concerned, we see the paragraph two as well as the possible exclusion and inclusion of certain systems as a result of further modifications in the text. Then on the issue of operator versus user, we are, I think, We don't have a preference in terms of whether it should be operator or user for us. What is important is to retain the word human in there. And that has been, I think, highlighted by in the intervention of the delegation from Ukraine as well, that we want to make it very clear that the word human is there and we are talking about further human intervention. We note that in the previous iterations of the definition, we did not have the phrase in the execution of these tasks. But then, of course, at that moment in time, we had a simpler formulation, which had noted without further human intervention. For now, I think what is there and keeping into account what has been introduced in the beginning of the text, the technological components and one or more weapons, I think it's important to retain what you have produced in the text. Lastly, Mr. Chair, we would just like to echo the proposal made by the delegation of Austria that perhaps we can have a look again at the characterisation in the light of the information that is provided by delegations. If we want to exclude certain systems in the future, we can certainly look back and approach and revisit this characterization at a later stage of our work. But at this moment in time, it's important before approaching other boxes that we are as broad as possible to address the risks that we would need to address when we approach other sections of the text. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:21:46]:

Let me thank the distinguished representative of Pakistan for further explaining his position and sharing his reflections. Next on my list is Ireland. Sir, you have the floor.


Ireland [1:22:03]:

Thank you for giving me the floor again on this topic. I just wanted to come in to echo some colleagues, including Austria, Pakistan, Brazil, and others. and respond to some of the clarifications on the rationale for preferring a cumulative definition and to reiterate why it's necessary to maintain a broad characterization at this stage. We also heard some arguments that seem to even further restrict the characterization to the point to we seem to be describing what we might call fully autonomous weapon systems. And for us, this goes too far for the purposes of what we're trying to do in this exercise. As some colleagues have said, this working characterization should also include systems that exist further down the spectrum of autonomy, which might fall under the regulation tier of the two-tier approach, such as loitering munitions. The cumulative approach and other further restrictions, such as including, for example, self-learning capabilities, is too narrow a definition for the purposes of the two-tier approach. The characterization should be broad enough to include systems that will be either prohibited or regulated or excluded entirely as per the second paragraph. We heard some delegations make references to specific weapon systems that currently exist that should be excluded from this characterization, and that's why we have that exclusion paragraph. But for the first paragraph, it need not be too overly narrow in order to avoid hamstringing our negotiation efforts later on. On lethality, we won't repeat our position here. I think it'll be useful to have some of the text in front of us, including from the proposals from the US and the ICRC, as we clearly have some way to go on that. And lastly, just to join others in maintaining the word human in the last line, again to underline the importance of the human element in our exercise. Thank you, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:23:54]:

I thank the distinguished representative of Ireland. I think this is it with regard to remarks on box one. Let me try to summarise, not summarise already results or provisional consensus, but summarise further work to be done by the chair and his team in the coming 48 hours or something with regard to a further review of the text in the second-half of the week. But let me start by really, really thanking you for your efforts to explain to the Chair, but also to all your colleagues in the room, other delegations, why you like certain things and why you don't like certain things, why you would like to change certain things, because that's helpful and we need that. And now, My takeaways from today's discussion, I would say we have had a very, very intense discussion on the scope of the working characterization. Let me repeat the words working characterization. I would like to refer also then to the remarks made by my distinguished Chinese colleague now a year ago with regard to the working characterization, that it is indeed not yet to define something that will be put into stone, but something that will provide us with a working basis, with a basis for further work with regard to the set of elements or other measures, blah, blah, blah, you know, the full mandate of the group, and I have the full mandate of the group, of course, in front of my eyes. So we should really have that. I would also be tempted to say that if we get it perfect in para one, we shouldn't have a para two. But we already assume we won't be perfect in para one, and that's why we have a para two. In para one, we have something that is broader, broader than a number of us would like it to have, and for that reason we have para two again to narrow it down or to exclude certain things. So we should not forget about that. Another thing is, of course, that this is, as was said also by a number in the room, the characterisation, and this is not yet clearly prohibitions and regulations. That is something for later in the text to look into. And whether they are legally binding, that is something that is not even in our mandate. That is something to be addressed in the next mandate. to be agreed by the high contracting parties at the review conference in 2026. So don't try to do that. That's not our mandate. Our mandate is already challenging enough, so stay away from that. Having said that, then we have the two main issues from today, I would say. That is indeed the issue around lethality and the issue around cumulative having cumulative tasks. Both, of course, pertain also to how broad the working characterisation is, and we should keep that in mind. I think, again, let me quote the Norwegian delegate, it should neither be too broad nor too narrow, and there we have to strike a balance. So if we work on the text on both Issues in the same direction that will have the strongest possible effect on, um, on the result, on the, on the, on the scope of the working definition, so one way or the other, we will have to balance it. Yes, I especially if I wake up early in the morning, sometimes I have certain remembrances of our attempt to have a footnote. We are not going to do that, but I think there is a basis in this room to work on what was called a one piece. And then, of course, we have to agree what would have to be in the one piece. We have references made to a proposal from the US, and I think once we have received it, we will start carefully studying it. We have also the proposal from the ICRC that was mentioned by also a large number of delegations in the room. So we have two competing text proposals, and we have to look clearly into the merits of both of them. and how to deal with them. Then there are what I would call the lesser problems, but they are still, of course, they won't go away out of their own, the issue of how to deal with integrated combination, if I may call it like that, but I think there have been some constructive proposals already. I think we also agree on what should be there. that it in a way is not trying to narrow it down. So I think we can find some solutions for that and we will come up with some solutions for that. And then of course we have had some suggestions to add language that is sometimes or that could be an alternative text, alternative version of what we have already, like further intervention in the execution or in relation to in the execution of these tasks, we have to look into that. Certain things were proposed in the past as a possible solution. If they are no longer considered to be a solution, I'm always happy to put them overboard. But there are also delegations in this room who are still attached to it, so we would then need to convince them that we can do without. I'm now looking at my notes and of my staff. Yes, with regard also to integrated combination or I think it was said by one delegation that it is about functional integration, so I think we have to look into that aspect with regard to the text. Yes, this is what I have, this is what are my takeaways. I do hope I have not missed some important issues you have brought forward during this very rich discussion, because I need, of course, to keep in mind all possible red lines from delegations, because we have to move towards consensus. I'm fully aware of that. But I think this discussion has been very helpful to better understand positions, and now we have the job to try and come up with language that would indeed reflect a convergence of positions. and we will try to do that over the coming 48 hours, as I said already, and present that to you for discussion later this week. So you are not done yet with regard to this. This is only step one and we will continue later this week. I am now going to look into my speaking notes with regards to box two, because I am not going to send you home, because we are going to make best use of the time we have and of the attention of the interpreters. So with regard to the second box of the rolling text. I would say that the key issue in this box is indeed human control, and human control is something we already talked about a lot and we will continue talking about it for a long time, because it will be in different places in the text, and I think that's something that was also said before by delegations. We have then to be consistent in our approach to that. And we have in this So far, we have had a number of suggestions with regard to the qualification with regard to human control, and also this was already addressed earlier today with regard to the working characterisation, but it will be again an issue in box number two. We have had context appropriate human control and judgement or meaningful instead of context appropriate while some of you still preferred not to include judgment at all or any other qualifier with regard to human control my understanding is that most if not all delegations agree on the concept or idea behind this specific wording but there are still divergences on how to capture that in concrete wording i have heard no positions that human control should not exist. So do we, for instance, agree that the term human control does not necessarily imply direct or real time and strict oversight of a human operator during all stages of use? Could it at certain moments be more indirect as appropriate for specific context, as is the case with certain already existing weapon systems? And there we get also an issue that we have been addressing already for quite some time today, that we are not only looking at the text, we are also having in the back of our mind concerns with regard to already existing systems or systems in development which our armed forces consider to be necessary and which we and which lawyers in respective capitals consider to be in line with IHL. So, Without further ado, I would now like to open the floor and invite delegations to provide their observations and possible proposals to bring about further convergence with regard to the text as it is now in front of us. For box two, and yes, we still have no sub, we still have no headings, uh, because I, that's another conclusion I've, I forgot to draw, we still have different positions on that, um, and I think we have already enough divergences. I'm not really tempted to start that discussion now on what should be there, perhaps with the exception that we could call box one the working characterization because I think we are there already at consensus level. But that's all from my side. I would now like to open the floor for delegations to share with us their views with regard to box two. We had, yeah, we had, since I think Brazil already kicked off earlier, let's now offer the opportunity to Turkey to kick off, to be followed by Brazil. Thank you for doing that, because I was on the point of assuming that this text was already consensual. But I think we need some further discussion and I'm happy to hear your views. The first opportunity is then for the delegation of Turkey. Sir, you have the floor.


Türkiye [1:37:43]:

Thank you, Mr. Chair. Actually, I'm taking the floor to Ayse Birik. Regarding section two, we note that the first three paragraphs reflect consensus language from previous DG reports. And the fourth paragraph is taken from the CCW preamble concerning the Martens Clause, albeit with the addition of the reference to laws. And for the fifth paragraph, we can join the consensus as it currently stands. Thank you very much.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:38:16]:

Thank you for that. Let me then pass the floor to the distinguished representative of Brazil. You have the floor.


Brazil [1:38:24]:

Thank you, Mr. Chair. My delegation fully supports the language contained in this box, which we believe offers the basic structure for a possible preamble section of a future instrument on the regulation of autonomous weapons. Our only proposal on this section would be the replacements in paragraph five of the word employment with the terms development and use. And we believe that the same expression should be used in other sections of the text, in particular boxes three, four and five. Mr. Chairman, we believe turning now to the formulation context appropriate human control and judgment, my delegation believes that it offers a lot of potential in terms of bridging the divides among delegations about the central notion of human control. In our view, the inclusion of the words context appropriate makes it clear that given the multiple situations and environments in which autonomy is to be employed in weapon systems, the technical requirements for human judgment and control cannot be subject to a one-size-fits-all approach. The formulation makes it clear that both judgment and control are mutually interdependent. The broader and subjective concept of human judgment refers mainly to the ability of individuals within the chain of command and control, commanders, supervisors, and operators, to take decisions related to the deployment of the weapon, taking into account the military doctrines, the different modes of operation, rules of engagement, training, and combat contexts. However, lawful AWS operations cannot rely only on human judgment or intent, essentially subjective contexts, concepts, sorry, but on the objective concept of human control over the critical functions and supervision to correct autonomous decisions that can produce collateral damage, override system failures or misinterpretations of the environment, targets, timing, and to achieve the desired outcome, both in military and legal terms. Thank you very much, Mr. Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:40:48]:

I would like to thank the distinguished representative of Brazil for his views. I have already quite a list of further requests for the floor, starting with Pakistan, to be followed by Ukraine, the ICRC, Sri Lanka and Switzerland. But let me first pass the floor to the distinguished representative of Pakistan. Sir, you have the floor.


Pakistan [1:41:11]:

Thank you, Chair. We would like to offer three comments in box two. First, relating to paragraph four, as we have noted earlier that we believe this box contains important foundational principles regarding our understanding of how IHL applies and our reaffirmation that IHL applies in cases of laws. We note that paragraph four is borrowed from the language contained in the preamble of the framework, the CCW framework agreement. So as of now, we believe that this language should be retained, of course, but when we read the language, we note that it gives an impression as if there would be some lethal autonomous weapon systems which will not be covered by the CCW, and we believe that debate is yet to be settled. So in that context, we would like to propose replacing laws with weapons in cases involving weapons not covered by the CCW, or generally just reproducing the language that is coming from the preambular portion of the convention. Then our second comment is on paragraph five regarding human control and judgment. we can go along with the language as suggested by you. For us, the notion of human control is particularly important. And when we look at how we arrive at this formulation, we note that in 2019, we had already agreed that human judgment was essential in order to uphold compliance with IHL, specifically in paragraph 17 of the 2019 report. Then in 2023, in paragraph 21, we also agreed that control is also essential in the context of upholding compliance with IHL. We believe that building on those significant agreements, it is important now to state those conclusions in one place, and our foundational principle should be that human control and judgment with regard to the use and effects of laws is essential in order to uphold compliance with IHL. Then our third comment is we would like to support what Brazil has also said, replacing the word employment with development and use. And we have similar edits later on in different portions of the text as well. We believe that for the sake of consistency, We can certainly replace development and use in all places of the text. And this is something that we note that we have already agreed previously that IHL does not begin at the moment of use. It must be embedded from the very inception of a weapon system. And this was a reaffirmation in the 2019 guiding principles as well. Additionally, when we look at Article 36 of Additional Protocol I, It requires state to determine at the stage of development and acquisition whether a new weapon system can be used in compliance with IHL. So these obligations exist because compliance with IHL cannot be an afterthought after the development has taken place. It's not just an engineering process. It is where fundamental legal, ethical, and operational choices are made. For instance, when we were going to talk about harmful biases, Uh, we know that those would need to be addressed in the training data of AI empowered targeting systems. Similarly, concepts of predictability, reliability, and explainability are, would need to be built during the development stage of those systems. So therefore, our regulatory framework cannot be limited just to the moment of use of those weapon systems. Therefore, we continue to believe that development and use are intrinsically linked, and we would need to ensure how to govern those systems, that they are designed in a way that makes lawful use possible in the first place. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:45:39]:

Let me thank the distinguished representative of Pakistan for sharing his views. Indeed, It has already been brought up by a few delegations, the suggestion to replace employment by development and use. I would be interested to hear also the views of other delegations, so if you intend to make an intervention and you have a pertinent view on that, I would appreciate you addressing that issue also. Next on my list is Ukraine to be followed by the ICSC and Sri Lanka, but let me first pass the floor to the distinguished representative of Ukraine. Sir, you have the floor.


Ukraine · Technical expert [1:46:22]:

Mr. Chairman, first remark regarding fast reaction on this text. In my opinion, in item one, we should be remark to international human love, not only in not in limited. but limited in framework of international human law. Second, my vision that item three, it is more topic of this part because it is very important result our work in previous three years because this text in this sentence give clear approach to use international human law regarding in context lethal autonomous weapon systems. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:47:17]:

We thank the distinguished representative of Ukraine for sharing his views. Next on my list is the ICRC to be followed by Sri Lanka and Switzerland. But first, let me pass the floor to the distinguished representative of the ICRC. Madam, you have the floor.


ICRC [1:47:33]:

Thank you, Chair, for giving us the floor. We welcome the text in box two in its current form, which reaffirms the applicability of IHL to autonomous weapons. We also welcome the incorporation of the text of the Martens Clause in paragraph four, reinforcing the key role played by ethics in underpinning and driving development of IHL. But we do share the concern outlined just now by Pakistan in being unsure how laws could fall outside the scope, and we would support replacing laws here with weapons. Additionally, we support the recognition of the importance of human control and judgment. In answer to the question you've posed just now, we see this not as requiring direct or physical control of the weapon system at all times. Indeed, this would remove the system from being characterized as an autonomous weapon system. Instead, we consider that the relevant control must be over the effects of the weapon system at all times in line with IHL requirements. And this control over effects is brought about by the combination of prohibitions and restrictions that we'll consider in box three. So namely prohibiting certain systems such as those with effects that cannot be controlled. and restricting the operating environment of others to limit the number of possibilities to a reasonable level to allow a user to anticipate and control effects. On the point about development and use, echoing the comments made this afternoon by Brazil, and Pakistan throughout the text, various terms are used and we would standardize these to development and use as a clearer alternative to things like deployment or employment. As has been outlined very well by the distinguished delegate from Pakistan just now, Covering development as well as use would serve to support compliance with IHL obligations such as the requirement to respect and ensure respect for IHL, which applies at all times, not only in conflict, as well as the requirement that parties to an armed conflict take constant care to spare the civilian population, not only in attacks but throughout their military operations. Applying aspects such as the prohibitions that we find in box three to development as well as use reflects the need to ensure that weapons are not developed that would likely be incompatible with IHL, and it would provide greater clarity to developers and industry more broadly. In the context of the CCW specifically, prohibiting the development of such weapons can be seen as promoting the purposes of the instrument by supporting the prohibition on use. This is analogous to amended Protocol II Article 8, which in order to promote the purposes of that protocol prohibits the transfer of any mine the use of which is prohibited by this protocol. I thank you, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:50:47]:

Let me thank the distinguished representative of the ICRC for sharing her views and making comments. Next on my list is Sri Lanka, to be followed by Switzerland, the UK and China. But first, the distinguished representative of Sri Lanka, you have the floor.


Sri Lanka [1:51:08]:

Thank you, Mr. Chair, for giving me the floor. As this is the first time that my delegation is taking the floor, I would like to take this opportunity to extend appreciation of Sri Lanka to Ambassador Robert Indenbosch of the Netherlands, the Chair of the GTE, for his leadership and his team to continued commitment in carrying forward the discussions on the rolling text, which provides basis for advancing the work of the GTE in accordance with its mandate. My delegation has comments in particular with regard to box two, as well as on the overall rolling text. As a country that strongly advocates negotiating a legally binding instrument in laws, Sri Lanka recommends that the references related to full compliance to the principles of international humanitarian law Centrality of human control and responsible chain of command and control throughout the entire life cycle of the autonomous weapon systems should be reflected, retained and strengthened and should not be diluted in the text. With regard to box two and the relevant sections of the subsequent boxes, it is noted that the rolling text in different paragraphs uses inconsistent terms to refer development and use of laws. particularly the term employment in box two para five. Similarly, in the subsequent boxes, we note the inconsistent terms such as use, employ, deploy or use in the box three and deployment and operation in box five. Therefore, we recommend to use the term either develop and use or development and use of laws throughout the rolling text in a consistent manner, replacing the different terms which have no common understanding and agreement in the context of law related to development and use of weapons. Therefore Sri Lanka suggests that all different terms be replaced with the term either develop and use or development and use according to the context which is in border sense encompasses all the terms such as employee employment deploy deployment and operation of laws. We have made these suggestions electronically to the chair. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:53:35]:

Let me thank the distinguished representative of Sri Lanka, also indeed to give her views on development and news. Next on my list is Switzerland, to be followed by the UK, China, Russian Federation, Colombia and Panama. But first, Switzerland, you have the floor, sir.


Switzerland [1:53:54]:

Thank you very much, Mr. Chair. I can say that we are generally very happy with box two. Paragraphs one to four are basic and fundamental restatements of IHL or aspects that we have agreed on in the general principles a long time ago. So that is actually stating the norm and we fully support this. Those elements possibly could be part of a preamble apart as basic considerations, but we are very flexible on that. Let me turn to five, because I want to say a few things about context appropriate human control and judgment. For many, I think meaningful human control has for a long time been the preferred concept, and it has shaped our work in the GGE for many years, and it was appreciated, I think, by many delegations because it includes a positive moral implication somehow, but I think it's also been made clear by many delegations that they think it would be really difficult to operate and implement in the real world. So in this light, we think that context of appropriate human control and judgment is a positive step forward. And it's actually quite remarkable that over the years we have moved in that direction. We see the term as really as a pragmatic compromise which can accommodate the concerns of states that have emphasized the need for control, but at the same time, those that stress the importance of human judgment. And I think we there, we have this flexibility also with regard to the implementation while maintaining a clear safeguards, clear safeguards against the risks posed by autonomous weapon systems. Um, I think some will say that those are two fundamentally different concepts and it's difficult to mix. We think that's exactly the added value of putting them together. Yes, they are distinct concepts, but we see a complementarity and we think they can really work in tandem to ensure lawful and, and responsible use. Control refers to the ability of humans to influence the, the functioning of the system, including activation, supervision, deactivation if necessary, but I would say also think the understanding has developed in the GGE that control doesn't mean… Um, judgment on the other hand, uh, ensures that decisions regarding the use of force are made based on human reasoning and an operation, and an understanding of the operational environment. We think both elements are important to prevent autonomous weapons systems from operating in ways that could lead to violations of IHL or unpredictable behavior in the battlefield. And again, as also just the ICRC has said, we are of the view that this is about control over the effects of the system. We need to look at this in a way that can be implemented in the real world. That brings me to the qualifier context appropriate. I think that really reflects the fact that the level and the nature of human involvement and human control must really be adapted to different operational contexts, including the type of weapon, the environment of the use and, for instance, the potential presence of civilians, et cetera. I think this has been a learning effect over many years in the GGE. And it also corresponds to the different levels of human involvement that may be required depending on the operational scenario. And I think it's important to take a concept that, you know, remains applicable across a range of technologies and different use cases. So that's why we quite are quite supportive of this context appropriate as a as an important component of the concept. Last but not least, would like to echo others that have said that IHL does not begin with the use but actually from the inception of a system, as our colleague from Pakistan just eloquently put it. So we would also, like other delegations, support instead of employment the consistent use throughout the whole text of development and use or development deployment of use. Thank you so much.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [1:58:42]:

Many thanks to the distinguished representative of Switzerland for sharing his views and for highlighting also certain aspects of previous discussions. And yes, we have not, let's say, shared away from, shied away from copying earlier text because we thought that Things that were agreeable in the past might still be agreeable now. For that reason, I think that paragraphs 1 to 3 are indeed taken from previous texts and state views on international humanitarian law that are commonly agreed. Next on my list is the United Kingdom to be followed by China and the Russian Federation but first UK so you have the floor.


United Kingdom of Great Britain and Northern Ireland [1:59:42]:

Thank you Mr. Chair and I think the UK also thinks that part two is in a good place. I'd like to skip forwards firstly to paragraph five and the use of the term context appropriate human control and judgment. As this meeting will be aware the UK have been using the concept of context appropriate for some years now. We actually use the term context appropriate human involvement and obviously would prefer that formulation but on the basis that our concept of context appropriate human involvement is our method of ensuring meaningful human control the UK can accept the context appropriate human control and judgment in this context. Equally in paragraph five we recognise the suggestions to change employment to use, sorry development and use and the UK could also support that change. Moving back to paragraphs one to four or specifically paragraph four, we recognise that these are restatements of existing international humanitarian law.

Although to be fair.

When we were reading paragraph four there was a little bit of uncertainty within the UK delegation as whether or not that did actually refer to the Martens Clause and we appreciate some of the comments so far that seem to suggest that it does. However, in order to really make sure that that's clear, the UK would suggest that we would remove all of the text before the term civilian population, i.e. where it says in cases involving laws not covered by the CCW annex protocols or by other international agreements, remove that text, that would make it, that would nail it much more further than that is referring to the Martens Clause and not to any other arrangements, conversations or agreements in any other fora or areas.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:01:59]:

Thank you for that. As has been said before, what has been deleted can never be wrong. It is an interesting proposal, in line with some of the other proposals to make changes to the bill. We will certainly reflect on that. Next on my list is China to be followed by the Russian Federation, Colombia, Panama. But first, China. You have the floor, madam.


China [2:02:28]:

Thank you, Chair. I have two points. The first is technical. In paragraph two, the current text gives us an impression when reading is that AHO generally applies to international and non-international armed conflict. Various additional protocols apply to various armed conflict. We hope a distinction be made in the text in this regard. Second point, a general comment on paragraph three in box two. In principle, we concur with the expression in this paragraph. However, in the conference zone, we have a consensus that laws has posed new challenges in the application of IHL. The GGE's mandate is to follow the principles of IHL, to take into full account the features, essence, and the ways of employment of laws, and the context of its use. So further clarified the interpretation, the application, and the execution of IHL in this context. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:04:21]:

I would like to thank the distinguished representative of China for her intervention and proposals. Next on my list is Russia, the Russian Federation, to be followed by Colombia, Panama, India. But first, the distinguished representative of the Russian Federation.

You have the floor, sir.


Russian Federation [2:04:40]:

Thank you very much, Chairman. With respect, the Russian Federation, on the whole, does not have any big challenges in accepting the text you've proposed in section two, box two. In terms of the substance, we agree with the understandings that you've included in box two. However, the only question that we would like to raise and that concerns the substance in this box relates to paragraph five and the concept of relating to the context of human control and judgment, in other words, context appropriate human control and judgment. So, Chairperson, we understand that for a number of delegations this concept is, to one degree or another, a concept that has been approved at national level. However, For us, what is important is specifically the substance and the essence of these concepts, and in the context of the previous activities by this group also. From our viewpoint, within this group, the concept of human control specifically has been seen as fundamental. in the context of the characteristics of autonomy and ensuring compliance with the norms of international humanitarian law as applied to these lethal autonomous weapon systems. Now, as concerns the concept of human judgment, Now, as we had understood this previously, human judgment has been defined and positioned in many ways as a substantive part of human control. Because for us, it is clear that ensuring human control without any human judgment is really not possible. However, human judgment alone cannot ensure human control. So it is specifically in the light of this understanding that we considered this issue during previous sessions of this group. And that is why also we consider that instead of further complicating this topic, it would make sense to stick with one concept, that of ensuring human participation or interference in the activities of such weapons systems. And in our opinion, the concept of human control includes within itself all of those aspects and implies itself all of that which many of delegations have listed prior to us speaking. So this is why we would prefer to stay with replacing this very cumbersome and complicated definition with the simple collocation human control. The qualification context appropriate is also a collocation which we don't fully understand. It seems to us that adding context appropriate can only lead us into additional discussions about how we are to understand this phrase. what does context appropriate cover, what does it not cover, what does it imply, what human control would be considered appropriate to this that context, what would not be considered context appropriate. So instead of this, we would prefer to talk in general about ensuring human control as an integral requirement which must be applied with respect to these weapon systems. This is something that we did successfully during previous discussions by the Group. We have the corresponding guiding principles on human control where it is clearly indicated that such weapon systems must remain under human control. Now the remaining comments related to this section are more about editing and style and not directly related to substance. Staying with paragraph five, We would prefer for the language to be more general, talking about human control with regard to lethal autonomous weapon systems as a whole, without narrowing this to the use and effects of laws. Because, in our opinion, The human control mentioned here concerns the systems as a whole and not just specific aspects of how they might function. Going back up to paragraph two, the last part of two, we think it's more justified not to talk about norms and principles of international humanitarian law as opposed to IHL in general. And so we would say that these norms and principles of IHL are applicable or apply regardless of the military technology that is used in weapon systems. As regards paragraph 4, we agree with the delegation of the United Kingdom on the feasibility of suppressing the first part of the language. This language appears to us to be somewhat superfluous. In our opinion, paragraph 4 concerns all or any cases. of the application of or use of one or other systems of weapons, including autonomous weapon systems, and should not be restricted just to the cases that are not covered by the CCW. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:13:31]:

Let me thank the distinguished representative of the Russian Federation for sharing his views and making proposals. Next on my list is Colombia, to be followed by Panama, India, Singapore. But first, the distinguished representative of Colombia. Madam, you have the floor.


Colombia [2:13:48]:

Thank you, Chairman. On box two, we'd like to make one or two additional points. Firstly, on paragraph three, We would suggest that we include the reference to public international law because we think the general regime of international law should also apply to laws. We don't just need a restricted reference to international humanitarian law here, in our opinion. That would be more like the Other clause, now we have a reference somewhere else here to international law, but it's a little bit restricted. Our second comment is directed towards paragraph five. The Group 16 submitted Protocol 6, as you remember, and there instead of a context appropriate human control, we used meaningful human control. From listening to discussions, we understand why we've got context appropriate here, but we've also listened to some delegations who don't really understand what context appropriate means. We think that our group proposal is a good one. I think it's important that both documents consider that there must be human involvement in the taking of decisions relating to the use of force. That's the main point at issue here. If context appropriate is not the appropriate word, we would suggest meaningful human contact, which was in our proposed Protocol 6. We can also support the proposal made by Brazil and supported, I think, by Pakistan and the ICRC, changing employment to development and use. Just one additional comment, if I may, on paragraph 4 while I'm here. We seem to be restricting the context a little bit. Maybe we should adjust this a little bit at the beginning perhaps laws should be replaced by weapons in general. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:16:31]:

Let me thank the distinguished representative of Colombia for her intervention and suggestions. Next on my list is the distinguished representative of Panama, to be followed by India, but first the distinguished representative of Panama, you have the floor.


Panama [2:17:03]:

Thank you very much, Chairman. On box 2, we have a general comment we would like to make, and that is that the current wording seems to be taking us in what we think is the right direction, and we're generally speaking in agreement with the way that things have been worded. I'd just like to make a few comments on some of the other proposals that have been made. In paragraph 4, for example, we would like to support the proposal that was tabled by Pakistan, given that we think this instrument needs to cover all autonomous weapon systems. and we don't want to leave a legal gap, the suggestion that there are some automated weapons systems which are going to end up not being covered by this. Then in paragraph five, and this is very similar to what my colleague from Colombia said, Panama being a member country of the Group of 16, which submitted the proposal on Protocol 6 relating to autonomous weapon systems, has always defended the centrality, if you like, of human involvement in the taking of decisions relating to the use of force. Our preference here would be to refer to significant human control or meaningful human control. As we understand it, we are looking for a compromise solution here, and we, generally speaking, would believe that the most fundamental thing in this paragraph is that we maintain the concept of human control and judgement and its link to compliance with international law. At the same time, we would like to offer our support to the proposal made by the Brazilian delegation that the word employment be replaced by the words development and use. This is in order to be consistent with paragraph 3 in this same box 2 and throughout the rest of the document as well. We think that when dealing with the question of autonomous weapons systems, we should focus on their entire lifespan, not just on certain aspects, as might be the case otherwise. They need to be covered from the very beginning. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:19:49]:

Austria, you are also on the list, or is it still your old place? You are on the list. So we have now first the distinguished representative of India. You have the floor.


India [2:20:05]:

Thank you, Chair. We believe that the delegation of decision making around the use of force to machines does raise legitimate legal and ethical challenges, especially around the question of responsibility and accountability in case of violations of international law, specifically international humanitarian law, which must continue to guide the conduct of armed conflict. Therefore, this section is especially vital and critical to the work of this group. India is supportive of the concepts and ideas contained in this section and is also broadly comfortable with the current wording proposed. Some limited specific textual edits have already been shared with your team and we can specify them later on in my intervention. However, let me take this opportunity to share our views on some of the debates and questions raised this afternoon by delegations. We've heard many delegations propose that we use the phraseology development and use of laws across the body of this text. Chair, while we completely endorse para three as contained in section two of the text, we do not believe that the same phraseology is relevant to other bodies of the other sections of the text which deal with regulations and prohibitions. Regulations, limits and controls must be limited to the use and effects of the system and not to the constituent technology and technological inputs. Therefore, the focus of prohibitions and regulations as are being considered in this rolling text should be on the actual operational use and the effects of the weapon system. This in no way dilutes the applicability of existing bodies of international law to the entire life cycle of the weapon system. including its design, development and deployment, as the group has already acknowledged in the guiding principle A of the GGE. However, it is our view that the focus of this group currently is to identify specific prohibitory and regulatory measures and elements that would ensure that laws are operated and used in a manner that does not violate international law and IHL. This is consistent in our view with how IHL applies to other weapons where prohibitions and regulations are linked to the effects and use of the systems and not on the development, unless specifically constrained by legal instruments or treaties. With regards to the question of applicability of IHL, we believe that international humanitarian law and the cardinal principles of distinction, proportionality and precaution contained therein apply to all means and methods of warfare, those of the past, present and of the future. These, in our view, would definitely include and be applicable to fully autonomous lethal weapon systems, which we believe are the subject of discussion in this GGE. We affirm our commitment to the continued application of relevant bodies of international law and believe that human judgment and oversight remain integral to ensure that the use of laws is done in a manner that meets the standards laid out under IHL. With respect para five and the wording around context appropriate human control and judgment, we have heard from distinguished delegates about the difference between control and judgment. In our view, both are subjective terms and it is not quite clear how these could be quantified. We believe that context appropriate human judgment is a broad enough umbrella term that contains within it ideas of command, control and oversight. And this would be our preference for the text as well. However, we do acknowledge that this remains a contested concept and we will need to do some more work as a group to arrive at a phraseology that appropriately includes these ideas and is acceptable to all. Lastly, with regards the wording in para 2 around armed conflict, we believe that the additional terms, whether international or non-international, are superfluous and may be deleted from the body of this text. Thank you, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:24:16]:

I would like to thank the distinguished representative of India for her intervention and suggestions. Next on my list is Singapore, to be followed by Austria, Ireland and Japan, but first the distinguished representative of Singapore. Sir, you have the floor.


Singapore [2:24:31]:

Thank you, Mr. Singapore appreciates that this section reflects our collective commitment to an international order based on the rule of international law, and in particular.

IHL in the context of this discussion, which applies wherever and whenever there is an armed conflict.

Jumping straight to paragraph five, we appreciate that the current formulation of the paragraph highlights the importance of a responsible chain of human command and control, per agreed language in the 11 guiding principles of 2019, while acknowledging the range of potential use cases of laws in modern conflict. At past sessions of the GGE, some delegations had expressed a preference to retain the notion of human-machine interaction per agreed language at the 2019 GGE, or the notion of human involvement in judgment. Some others are passionate about the use of the term meaningful human control. If one looks back at the final report of the 2018 GGE, there is a whole buffet list at page 16 of possible terminologies centred around the human that were deliberated previously.

In Singapore's view.

What matters is that the decisions over the level of human control and judgement required in the operation of laws.

Should always take into account, one, the lawfulness of its intended use, and two, contextual factors.

As with all things in life, context matters.

I thank you, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:25:58]:

I would like to thank the distinguished representative of Singapore. Next on my list is Austria to be followed by Ireland, Japan, Israel, USA. But first, the distinguished representative of Austria. Sir, you have the floor.


Austria [2:26:14]:

As others, we can state that POCTWO is in a very good shape and I will not make many comments that others have not already made. I also want to stress that you have been very diligent in quoting the existing language that we are having, including also having the different positioning of methods or means of warfare in para one, but we might even consider to switching so that we have some coherence with the other paras, but this is no hard point, just something that we could do as long as we all agree to do it, being fully aware that this is a quote from the preamble of the CCW. The second point that has already been made by many is that para four is an interesting attempt by taking the idea of the Martens clause and pushing it over or taking it over for the purpose that we are discussing here and for the purpose of laws. I think we support the idea that this should be kind of shortened but different from others. We would suggest that we take it directly as it is also in the preamble, which would also not include weapons but which speaks about cases. And I think there's an advantage here as this is broader and also relates to the idea of the Martens Clause, which does not only speaks about weapons but per se has a very broad application, which probably refers to means and methods of warfare in general. Having said this, I mean, there's probably some merit in the idea that you had, but this would probably require some kind of Martens Clause for the 21st century that goes into the risks and challenges by new technologies. But I think this is not something that there is much appetite in the room, but I will keep the idea in the back of my head and maybe at some point it will come out again. On paragraph 5, we share the two points that many other delegations have made. From our point of view, we still see merit in having the word 'meaningful' in there, meaningful human control, and the disqualifier helps to avoid the situations that we have described as nominal human input. This is something that is important. At some point in our discussions, we even had in the characterisation section. It has been lost. I think it's still important to come back to this question and to solve it in some part of the text or the other. Meaningful would be an elegant and known way to do this and this could be a place to do it again. So our preference would be context appropriate and meaningful human control. We could also be fine with judgment, but as others have already stated in our interpretation, control already includes judgment to a large extent. So it's not per se needed, but we are flexible on this field. Next point, we also support to switch out employment to development and use. The whole life cycle issue has been discussed also since a long time, and this is something we definitely support. It is important as what we are doing here is upholding IHL and IHL does not only apply to the use of force, but it's a broader endeavor and with new technologies, we need to be sure that the development stage and the actors that are active in this stage are involved in the whole exercise that we are doing. And this also makes life easier for states. because in the end they will be the ones to have to do the Article 36 review. So this is something that should be definitely included and has big merit and also usefulness for states. And I think that's everything from my side. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:30:05]:

Let me thank the distinguished representative of Austria for his intervention. Next on my list is Ireland to be followed by Japan. But first now the distinguished representative of Ireland. You have the floor.


Ireland [2:30:19]:

Thank you, Mr. Chair, and I can be brief. We fully support box two and echo many of the previous comments, including that throughout the text we see various terms that are used such as employment or deployment and use. And for clarity and consistency, and as well for the reasons eloquently outlined by our colleague from Pakistan, we believe that the terms development and use should be standardized throughout the text where appropriate. Due to the unique nature of the development phase of laws, including developing algorithms, IHL and other international law should be considered throughout its life cycle. Like others, we very much welcome paragraph five, including the essential aspects of context, appropriate human control and judgment with regard to the use and effects of laws. For us, this is a very clear and concise formulation that gets to the heart of what we're trying to do here. And we fully support its inclusion and join others as well in adding meaningful before human control, which further drives the point home. For us, the concept of human judgment is important and distinct from and complementary to human control. In the context of laws, it is essential to underline the importance of maintaining human judgment when ensuring compliance with IHL. And lastly, and this is mostly a semantic suggestion in paragraph five, which is perhaps a change in the second line, is essential in order to ensure to is necessary to ensure which makes it a bit more direct thank you chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:32:00]:

Let me thank the distinguished representative of Ireland for his views and concrete suggestions next on my list is Japan to be followed by Israel the USA Australia but first the distinguished representative of Japan sir you have the floor.


Japan [2:32:13]:

Thank you oh thank you Mr. chair Government of Japan is basically of the view that the obligations of international humanitarian law apply to all weapon systems, including laws, of course, and support the notion that weapons systems inconsistent with IHL should not be used and the necessary restrictions should be put in place to ensure compliance with IHL. Now, We support the idea of this technology neutral approach being deflected in this content of this box number two, and particularly as described at the end of the paragraph two. Now, my government is now of the view that limiting the technology covered by this box to military use only may create some loopholes in the application of international humanitarian law in light of the fact that many technologies have both military and civilian uses at this moment. So regardless of this, whether the technology is military or civilian or dual use, if it is used in military context, so it could be subject to discussion regarding these laws. So My delegation's preference is paragraph 2, the last part is applicable independent of the technology used. Thank you very much, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:33:53]:

I would like to thank the distinguished representative of Japan for sharing his views and making proposals with regard to the text in front of us. Next on my list is the distinguished representative of Israel, to be followed by the US and Australia, but first Israel. You have the floor, sir.


Israel [2:34:11]:

Thank you so much, Chair. Regarding section 2, our position is that the fifth paragraph should be deleted. I will explain. We strongly object the language context of "appropriate human control and judgment". The term "human control" has been extensively discussed in the GGE meetings and due to its highly controversial nature, sorry, due to its highly controversial nature and was not adopted in any consensus outcome as the appropriate description of the human-machine interaction. Our opinion on this term and its substantive problems has remained the same. The addition of the words "and judgment" further exacerbates the problem rather than resolving it. The context appropriate element can easily be understood as referring to an obligation of constant real time control, which is in the heart of our debate. For these reasons, we cannot accept the suggested terminology. We will suggest using language that has already achieved consensus, such as appropriate human involvement. In addition, regardless of any textual change, we think that this paragraph should not be appearing in part two of the paper, since this part is focused on fundamental issues concerning the applicability of IHL rather than IHL obligations themselves, let alone elements which are not even existing obligations. Indeed, the issue of human machine interaction appears in later parts of the paper. Thank you, Chair.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:36:04]:

I would like to thank the distinguished representative of Israel for his views and also his comments. And indeed, we have in box four, paragraph two, text that is not the same, but that covers part of what we are discussing here, so there are other places in the text where we have it. But on the other hand, if you want to delete it fully, it is not an issue of moving it. But that takes some more views with regard to this box. I have the USA first, to be followed by Australia, Mexico, Finland and the Republic of Korea. And I have my doubts whether we are going to make that before six o'clock, but I think so far we have made good progress anyhow. But let's first listen to the United States. You have the floor, sir.


United States of America [2:37:04]:

Thank you, Mr. Chairman. Chair, we would like to, I think, focus our comments in this box on paragraphs two and five. So with respect to paragraph two, I'll suggest the edits that we're interested in and then explain why we're suggesting these edits. We would prefer to begin paragraph two with, by inserting four words, the cardinal principles of IHL apply to armed conflict, whether international or non-international. And then we would propose to move the clause that's at the end of this paragraph and place it after non-international. So it would read whether international or non-international, comma, independently of the specific technology used. And then we would insert consequently governs the use of all weapons, et cetera, to the end of the paragraph. And just to explain why we're proposing this. So the insertion of the cardinal principles is a technical point. And I think the technical point that we're trying to capture here is that some IHL rules depend on the character of the conflict or on the particular weapons or means or methods of warfare used. And so The ICJ's advisory opinion, which I think actually inspired this paragraph, itself uses the formulation, the cardinal principles of. We would be bringing it, I think, more in line with the way the ICJ formulated its own opinion. And so again, I think this is intended to be a more precise formulation. on moving the clause, it just seemed to us as a matter of logic that placing it in this position would just be a little bit clearer. And what we also did was we replaced the word military with specific, so independently of the specific technology used. And the legal principles here apply even if the military technology isn't used. For example, IHL applies if a specific technology used in armed conflict was developed by civilians or by terrorist groups. We just want to have more of a focus on the technology. Chair, with respect to paragraph 5, I think we have much more fundamental concerns here. First, I think some delegations have noted that What this paragraph is attempting to do is to merge constructs that we've discussed in the past, the concept of control and the concept of human judgment, and place it into a singular sentence. I think as a starting point, we see these as two very different things. I think a number of delegations have already intervened to explain the differences. And historically in this body, we've dealt with them in different places and we've approached them differently. I think the two most important agreed provisions on these are found in paragraph 21 of the 2023 GGE report as it relates to control. and paragraph 17E of the 2019 GGE report as it relates to human judgment. So for us, I think it's important to remind delegations, as we have in our working papers and in our prior statements in this GGE, that we have very, very serious concerns with references to human control. We find it to be overly narrow. We don't think it's a good way of describing the relevant issues and the challenges that face us. IHL itself does not require human control as such, and software systems may lawfully be used to control autonomous weapons systems. We have not had a guiding principle or any other agreed reference to human control in the past. I think some delegations have said that, but it's not the case. And so, so we think it's really important for us to continue to stress our concerns with this. We also have a structural concern with how the term is being used in the document. As we've described before, we have a serious concern, and I think it's been illustrated throughout the interventions this afternoon, with the idea of trying to approach this by adopting a new, vague, overarching standard like meaningful or context appropriate human control. And we've listened very carefully this afternoon to the interventions, and we continue to hear very large differences in the way that delegations describe what they mean by meaningful human control or context appropriate human control. And throughout the years that we've been having this discussion, what we've heard, and I think we heard it in the ICRC intervention as well today, that they would want to anchor the idea here, but then to describe what is relevant about this concept later on in box three. I think our basic approach has been to try to move away from this label. It hasn't found consensus. We don't think it will find consensus. And to the extent that we are focused on specific measures in box three and elsewhere, that we get to work and we identify those measures. Identifying a concept and then saying that these measures flow from the concept to us is a misunderstanding of what we're trying to achieve here. We're also concerned that this approach will inevitably create confusion and potential conflicts with the existing standards in IHL that already apply. And so, again, throughout this debate, we have really tried to focus first and foremost on articulating existing IHL standards. and then after that exercise to consider what further standards, if any, we think should be adopted. Chair, you asked for reactions on replacing references to use with references to deployment and use or -- sorry, development and deployment or development and use throughout the document. And I think people have basically said, why don't we just do a find and replace throughout the document. It's true that we've incorporated references to development in the past. For example, in guiding principle A, we do have a reference to development. And I think the formulation in paragraph three in this box corresponds closely to what we agreed in guiding principle A. But we're not in a position at this stage to support the broad replacement of references to use with references to both development and use. In our view, IHL is principally concerned with use. We don't accept, for example, the ICR interpretation that the norm of respecting and ensuring respect with the 1949 Geneva Conventions has created a customary international law obligation that is so broad as to encompass this type of obligation with respect to development. So we agree that it's critical and necessary for states to take measures in development that ensure that weapons can be used in compliance with IHL during armed conflict. The proposal that we and a number of our co-sponsors have submitted have some language to this effect, but I think it's primarily, in our view, a matter of effective implementation. And a violation of IHL doesn't occur if during development a particular measure is omitted. Rather, the violation of IHL occurs when the weapon is used in violation of IHL in armed conflict. And by the way, the The way we would approach paragraph five then to sort of bring all of this together is to suggest edits that would, if we're going to merge what we see as paragraph 21 in the 2023 GGE report and paragraph 17E of the 2019 report, we would want to see some textual adjustments. By the way, we would be perfectly happy to just retain those two paragraphs and to replace what is currently paragraph five with those two paragraphs. But if we're going to merge them, we think they should be merged in a way that remains true to those two paragraphs. And so our edit here would delete context appropriate human. It would begin the paragraph with control and human judgment. with regard to laws are needed to uphold compliance with IHL, including the principles and requirements of distinction, proportionality, and precautions in attack. That is, in our view, a truer reflection of paragraph 21 of the 2023 report, which read, 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. This reformulation, we think, would also avoid some of the discussion we've just had on the question of development. So I'll stop there, Mr. Chairman. Thank you.


Netherlands (Kingdom of the) · Chair · Robert Indenbosch [2:48:20]:

Thank you for that and certainly something to further reflect on. We are at the end of the interpretation and let me really thank the interpreters for a great job today. I fully understand this has not been easy. It's very technical, it's very specific, but you are of great help to better understanding each other's positions. Tomorrow, I have still one, two, three, four, um, um, nation, uh, high contracting parties on my list being Australia, Mexico, Finland, and the Republic of Korea. So we start tomorrow, um, we discuss, uh, continuing the discussion on box two. I would now like to invite you to some refreshments and they are served one floor up. So it's not as easy as it was other times when we were just in front of 19. You now have to move one floor up and there you will find some drinks and some snacks. So you will be able to continue the informal discussion. Very much looking forward. And that was all for today and see you tomorrow morning.