Shaky red lines? Anthropic pitched Pentagon autonomous drone swarm technology

Anthropic reportedly pitched the Pentagon technology to coordinate an autonomous drone fleet amid fraught negotiations over the military uses of artificial intelligence (AI).

The maker of Claude competed in a $100 million Pentagon prize challenge to create voice-controlled, autonomous drone swarming, at the same time that the company was clashing with the Department of War (DoW) over its red lines, Bloomberg reports.

Secretary of War Pete Hegseth labeled Anthropic a “supply chain risk” on February 28th after the company refused to yield to the Pentagon’s demands to remove its guardrails.

Anthropic cited two “red lines” – its technology being used for “domestic mass surveillance” and “autonomous weapon systems.” The Pentagon said it would use AI tools for “all lawful purposes.”

Anthropic’s refusal prompted President Donald Trump, who previously criticized the company for being “woke,” to ban its tools across government agencies.

According to a source familiar with the matter, Anthropic didn’t think developing the technology would cross its red line, as a human would still be able to monitor and stop lethal drone swarms.

Its submission focused on using Claude to translate a commander’s intent into digital instructions and to coordinate a fleet of drones, instead of using AI for autonomous targeting or weapons decisions, Bloomberg said, citing an anonymous source.

Anthropic wasn’t among the companies selected for the drone fleet, according to people with knowledge of the matter.

The company declined Bloomberg’s request for a comment.

Cybernews has reached out to Anthropic for comment and will update this article once it is received.

Anthropic’s decision to draw red lines – and lose a $200M contract with the Pentagon – received a lot of praise on social media, with European activists calling for the company to move overseas.

However, Claude was reportedly used in the US attack on Iran hours after the Pentagon blacklisted the company.

The company’s CEO, Dario Amodei, has repeatedly said he was hoping they could still reach an agreement with the DoW. Unsurprisingly, some AI security researchers remain skeptical of Anthropic’s newfound focus on limiting the military use of AI.

Timnit Gebru, the founder and executive director of the Distributed AI Institute, emphasized the discrepancy between Anthropic’s participation in the Pentagon’s competition and praise for being “ethical.”

Luiza Jarovsky, co-founder of the AI, Tech & Privacy Academy, wrote on X that Anthropic and OpenAI, which recently unveiled a deal with the Pentagon, “share very similar views on the use of AI for military purposes.”

Jarovsky wrote, “Their main difference is a governance one: Anthropic wants the final word on what’s right and what is wrong regarding how AI affects people and society. It doesn’t want to be bound by the legal system.”

Unlock more exclusive Cybernews content on YouTube.