New draft report - Statement by Stop Killer Robots to the GGE on LAWS, 31 August - 4 September

The following statement was delivered on Thursday, 3 September 2026, by Elizabeth Minor to delegates participating in the meeting of the Group of Governmental Experts on Lethal Autonomous Weapons Systems (LAWS), which took place on 31 August to 4 September 2026.

Thank you, Chair.

Stop Killer Robots would like to again express our appreciation for the efforts of you and your team to attempt to preserve the work of the group in the set of elements text, whilst attempting to reach consensus – and in the context of discussions this week that have involved the reopening of issues that have already been discussed extensively, as well as last minute attempts to make fundamental changes, after years of developing common understandings and making compromises.

Overall, we believe that the text in section III A of the revised draft report still preserves the main points of progress achieved by the group, in terms of an adequate characterisation of autonomous weapons systems, and the elaboration of specific measures for exercising meaningful human control, as well as the core recognition that control and judgment are activities undertaken by people, and that control is exercised by human beings. 

We would like to echo the comments of delegations who have observed that this text is not optimal, for reasons we have outlined before; that it should represent a minimum towards moving to the next step; and that the package as a whole must now be kept and built on.

To briefly make some specific points as we will not be present at the informals: regarding paragraphs 35 to 37, we are deeply concerned that the text could now be read as suggesting that the measures outlined in paragraph 37 and elsewhere in the document should be an optional menu of measures – rather than being a baseline set of ways in which human judgment and control can be achieved, among other measures, and which when negotiated further and turned into legal rules, states must implement. 

Indeed, we imagine this may be the intention of the changes for some. 

It is also concerning in this regard to see the words “as feasible” added to paragraph 36.

And within the package as a whole, we also do not see why the concept of “explainability” has been removed from 39(c). Though we welcome that predictability, reliability, and traceability have been retained, explainability has always been part of the package of measures that are essential to ensuring that systems are sufficiently functionally understood.

At this point, when states undertake the next stage of their work – of using their discussions on these elements as a baseline for negotiating a strong international legal instrument on autonomous weapons systems – they must ensure that: all essential measures to ensure meaningful human control are obligations, not options; that systems that cannot be used in accordance with these measures for control are prohibited; and that anti-personnel autonomous weapons are also prohibited.

Thank you, Chair.