More than 200 civil society groups and stakeholders on Monday required an immediate halt to the use of artificial intelligence systems in military “kill chains” and warns that AI-accelerated warfare risks encouraging violations of international criminal law, human rights and humanitarian law.
The statement warned that AI systems embedded in killing chains are accelerating the speed and scale of military attacks in ways that undermine fundamental principles of humanitarian law, including distinction, proportionality and caution. The signatories said that supposed safeguards such as “human in the loop” mechanisms cannot prevent the deadly consequences of AI-accelerated targeting, but instead risk becoming a means of “sanctioning” killings at greatly accelerated speed and scale.
The coalition pointed this out US and Israeli attacks on Iran As an example earlier this year, it noted that AI target generation tools enabled attacks on nearly 2,000 targets within the first 48 hours of the campaign. In Gaza, Israel used AI targeting tools, including Lavender, Gospel and Where’s Daddywhich the statement said could “contribute to the concealment of international crimes behind the guise of perceived algorithmic objectivity.”
Several major technology companies were named in the statement. OpenAI has agreed to provide AI services to the U.S. Department of Defense, and Google has contracted with the Department of Defense to develop prototype frontier AI capabilities for warfare. Anthropic’s major language model Claude reportedly played a role in supporting US attacks on Iran. The statement quoted a Standoff between Anthropic and the US government about military use, and noted this more than 560 Google employees signed an open letter in April calling on the company to reject the use of secret military AI.
The signatories called on technology companies to refrain from military contracts with agencies that violate international law and to discontinue AI decision support systems for targeting. They called on states to stop using AI on military targets and provide transparency about how it is currently used in combat. The statement noted that Amnesty International has written to both OpenAI and Anthropic about their human rights policies regarding generative AI in a military context; At the time of publication, only OpenAI had responded.
