Pentagon wants killer AI without safeguards – Reuters

The US Department of War has reportedly clashed with contractor Anthropic over ethical limitations built into its tech

The US Department of War is locked in a dispute with artificial-intelligence developer Anthropic over restrictions that would limit how the military can deploy AI systems, including for autonomous weapons targeting and domestic surveillance.

The disagreement has stalled a contract worth up to $200 million, as military officials push back against what they see as excessive limits imposed by the San Francisco-based company on the use of its technology, Reuters reported, citing six people familiar with the matter.

Anthropic has raised concerns that its AI tools could be used to carry out lethal operations without sufficient human oversight or to surveil Americans, sources told Reuters.

Pentagon officials, however, have argued that commercial AI systems should be deployable for military purposes regardless of a company’s internal usage policies, as long as they comply with US law.

Read more

The dispute comes amid a broader push by the Trump administration to rapidly integrate artificial intelligence across the armed forces. Earlier this month, the Department of War outlined a new strategy aimed at transforming the US military into an “AI-first” fighting force.

The Pentagon believes it must retain full control over how AI tools are employed on the battlefield and in intelligence operations, with Defense Secretary Pete Hegseth vowing not to use models that “won’t allow you to fight wars.”

An Anthropic spokesperson said the company’s AI is “extensively used for national security missions by the US government” and that it remains in “productive discussions with the Department of War about ways to continue that work.” The Pentagon has yet to comment on the reported rift.

Read more

Anthropic chief executive Dario Amodei has repeatedly warned about the dangers of unconstrained AI use, particularly in mass surveillance and fully autonomous weapons systems. In a recent essay, he argued that AI should support national defence “in all ways except those which would make us more like our autocratic adversaries.”

The standoff poses risks for Anthropic, which has invested heavily in courting government and national-security clients and is preparing for a potential public offering. The company was one of several major AI developers, alongside OpenAI, Google and Elon Musk’s xAI, awarded Pentagon contracts last year.