8th October 2026
Our Stop Killer Robots UK panel event on the fringes of the Labour Party Conference in late September, chaired by campaign coordinator Andy Bailey, brought four experts together for a thoughtful conversation on next month’s CCW Review Conference and the UK’s position on treaty negotiations.
Dr Alexi Drew, Dr Matthew Mahmoudi, Professor Thompson Chengeta and Namir Shabibi set out how warfare is changing, what meaningful human control means in practice – and what we can do about it.
The Genie is out of the bottle
Autonomous weapons and AI-assisted targeting systems are already used in modern conflicts – “in the wild,” as one speaker put it – in Gaza, Ukraine and Iran. This is not a future sci-fi extinction threat, as leading tech experts have sought to frame the debate in recent weeks. Dr Matt Mahmoudi, researcher and advisor on AI and Human Rights at Amnesty International, explained how, “by playing into the annihilation events framework, we’re playing into the narrative of the tech companies themselves who want to be seated at the table in which they’re making decisions about safeguards.” Using the example of target acquisition systems like Lavender, Where’s Daddy and Gospel used by Israeli authorities against Palestinians in Gaza, which draw on mass surveillance data to establish targets, he emphasised the harms and existential threats already facing people today.
Panellists warned that trying to put this technology back is now impossible. Instead, there is an urgent responsibility to establish strict legal and technical constraints to govern its use – including at the Seventh Review Conference of the CCW in November – before it becomes completely normalised.
“Meaningful Human Control” versus “Human in the Loop”
Ensuring the compliance of weapons systems with International Humanitarian Law (IHL) requires the ability to stop or alter an action if it is deemed to breach compliance with key provisions and principles: namely precaution, distinction and military necessity. Placing a human “in” or “on the loop” of autonomous weapons systems and AI assisted tools is insufficient if the human cannot realistically exercise control, nor if such tools are blatantly already in breach of IHL obligations. The recent adoption of the GGE LAWS text in September also explicitly recognised that ‘control and human judgement’ are necessary to uphold international law.
Furthermore, AI processes data and generates information, including potential targets, at a pace that overwhelms human cognitive capacity. Over time, operators can develop uncritical trust in the system, known as ‘automation bias’, which reduces deeper interrogation of unlawful or unethical strikes and weakens oversight. Dr Alexi Drew, Director of technology consultancy Penumbra Analysis and former Technology Policy Advisor at ICRC, acknowledged the reality of where we find ourselves and the so-far missed opportunities, but emphasised how that “simply means we have a greater responsibility to legislate and constrain safely the thing that we have contributed to the creation of.”
The reason we use meaningful human control is because the concept of having just a human on the loop doesn’t necessarily mean that they have the ability to operate that function effectively.
Dr Alexi Drew
Director of Penumbra Analysis and former Technology Policy Advisor at ICRC
Techno-laundering and dehumanisation
‘The panel warned against “techno-laundering war crimes” – the risk of allowing military forces to blame ‘the machine’ for atrocities, rather than facing accountability themselves. “To focus on these systems and to repeat the narrative that ‘the machines made us do it’ is actually serving the IDF and Israeli government propaganda,” investigative journalist and researcher Namir Shabibi explained.
Shabibi spoke about how the UK Ministry of Defence (MOD) has pursued projects aimed at multiplying military lethality through digital targeting, and Mahmoudi dismantled claims that biometric and facial recognition technologies could reduce crime rates or bring casualties down, challenging the illusion of accuracy. Contributing to the panel’s conversation on the human cost of surveillance and automated targeting, Mahmoudi also referenced the computer science saying ‘Garbage in, garbage out’ and emphasised the difficulty of accurately assessing through artificial intelligence whether someone is actually a combatant, let alone determining under IHL how that decision was made, or by whom.
We see overwhelming evidence that these technologies tend to lead to a digital form of dehumanisation, more so than anything else.
Dr Matt Mahmoudi
Researcher and Advisor on AI and Human Rights at Amnesty International
Tech sector accountability and what the UK must do next
Companies building military AI components and tools, including related infrastructure such as datacentres, also face responsibility. Drew used the example of the US-Israeli war on Iran, where companies that might not see themselves as direct participants in hostilities have been seen by other participants as a target. Panellists noted how these entities are facing consequences, including being treated as targets or facing legal challenges under international frameworks, and that this will only increase.
The UK has agreed with UN resolutions on the risks of autonomous weapons, but continues to resist backing a legally binding international treaty. Panellists agreed that ahead of the crucial November review conference, the UK must support a legally binding instrument that establishes clear limits.
Written by Alannah Travers