Anthropic boss rejects Pentagon demand to drop AI safeguards

ReutersAnthropic chief executive Dario Amodei

Anthropic has said it will not back down in a fight with the US Department of Defense (DoD) over how its artificial intelligence (AI) technology is used.

The firm's chief executive Dario Amodei said on Thursday that his company would rather not work with the Pentagon than agree to uses of its tech that may "undermine, rather than defend, democratic values."

His comments come two days after a meeting with US Secretary of Defense Pete Hegseth over demands that Anthropic accept "any lawful use" of its tools. It ended with a threat to remove Anthropic from the DoD's supply chain.

"These threats do not change our position: we cannot in good conscience accede to their request," Amodei said.

At issue for Anthropic is the potential use of its AI tools like Claude for two purposes: "Mass domestic surveillance" and "Fully autonomous weapons."

Amodei said "such use cases have never been included in our contracts with the Department of War, and we believe they should not be included now."

The Department of War is a secondary name for the Defense Department under an executive order signed by US President Donald Trump in September.

"Should the Department choose to offboard Anthropic, we will work to enable a smooth transition to another provider," Amodei said.

An Anthropic spokeswoman added on Thursday that while the company received updated wording from the DoD for its contract on Wednesday night, it represented "virtually no progress on preventing Claude's use for mass surveillance of Americans or in fully autonomous weapons."

"New language framed as compromise was paired with legalese that would allow those safeguards to be disregarded at will," she said. "Despite [the Department of War's] recent public statements, these narrow safeguards have been the crux of our negotiations for months."

A representative of the Defense Department could not be reached for comment.

Before Amodei's comments, Sean Parnell, a spokesman for the Pentagon, wrote on X, "This narrative is fake and being peddled by leftists in the media", referring to claims the DoD wanted to use Anthropic for mass surveillance or fully autonomous weapons.

"Here's what we're asking: Allow the Pentagon to use Anthropic's model for all lawful purposes", Parnell added.

A Pentagon official previously told the BBC that should Anthropic not comply, Hegseth would ensure the Defense Production Act was invoked on the company.

The act essentially gives a US president the authority to deem a given company or its product so important that the government can require it to meet defence needs.

Outsourcing IT services.anthropic boss rejects

But Hegseth also threatened to label Anthropic a "supply chain risk", meaning the company would be designated as not secure enough for government use.

A former DoD official who asked not to be named told the BBC on Thursday that Hegseth's grounds for either measure were "extremely flimsy".

A person familiar with the negotiations who asked not to be named said tensions between Anthropic and the Pentagon "go back several months," before it was publicly known that Claude was used as part of a US operation to seize Venezuelan President Nicolás Maduro.

While Amodei did not specify exactly how Anthropic could be or had been used by the DoD for mass surveillance or fully autonomous weapons, he wrote in a company blog post that AI can be used to "assemble scattered, individually innocuous data into a comprehensive picture of any person's life – automatically and at massive scale."

"We support the use of AI for lawful foreign intelligence and counterintelligence missions," Amodei said. "But using these systems for mass domestic surveillance is incompatible with democratic values."

As for AI being used in weapons, Amodei said even today's most advanced and capable AI systems "are simply not reliable enough to power fully autonomous weapons."

"We will not knowingly provide a product that puts America's warfighters and civilians at risk," Amodei said. "Without proper oversight, fully autonomous weapons cannot be relied upon to exercise the critical judgment that our highly trained, professional troops exhibit every day. They need to be deployed with proper guardrails, which don't exist today."

He added that Anthropic had "offered to work directly with the Department of War on R&D to improve the reliability of these systems, but they have not accepted this offer."

Hegseth had demanded the Tuesday meeting with Amodei, a source previously told the BBC.

Artificial intelligenceDonald TrumpInternational business

Original aticle here: BBC