Anthropic and Google Named in 130-Group AI Warfare Halt Demand

A joint statement published by Access Now and signed by over 130 organizations and 165 individuals does something that most AI ethics statements avoid: it names names. According to the statement, citing an NBC investigation, AI-enabled systems including Anthropic's Claude large language model and the Maven Smart System are playing a role in supporting US and Israeli attacks on Iran. The statement also references Israeli systems called Lavender, Gospel, and Where's Daddy as used in targeting in Gaza.

The core argument is that "AI-accelerated warfare is rapidly becoming a means of rubber-stamping killing at speed and at scale, and currently no technical or procedural fixes can effectively prevent the lethal and devastating consequences that stem from the fundamental challenges it poses to international law." That framing directly rejects the industry-standard defense of "human in the loop" oversight and hardcoded guardrails as inadequate safeguards.

The signatories assert that the use of LLMs for target generation and prioritization pushes military actors toward warfare where foundational principles of international humanitarian law, including distinction, proportionality, and precaution, cannot be sufficiently respected given the speed and scale of the technology, and the "unreliable, biased, and often illegally obtained input data" feeding these systems.

Companies named alongside Anthropic include OpenAI, Google, Microsoft, and Amazon. The demands run to both tech companies and states: stop providing AI systems for military kill chains, and provide transparency on current military AI applications. The statement acknowledges a structural difficulty, noting that "once companies have entered military contracts, they may have limited agency over how their products and services are used," pointing to Anthropic's standoff with the US government and reports of Google and Amazon suspending the applicability of their terms of service in contracts with the Israeli government.

What the statement does not give you is a technical or legal definition of where kill-chain involvement begins and ends, or evidence that any of the named companies have committed to changing course. The 165 individual signatories reportedly include people from Google, OpenAI, and Microsoft, with the statement noting their signatures do not represent their employers' positions.