The US military could punish Anthropic for refusing to amend the lab’s ethics code, a report says

The US Department of War is close to cutting ties with key AI partner Anthropic, the company behind the Claude model, and designating it a supply chain risk, Axios reported on Sunday, citing a Pentagon official. The label is typically reserved for entities linked to states that Washington considers foreign adversaries.

The San Francisco-based lab has reportedly been clashing with the US government for months over its policy limiting the military use of its technology. While the Pentagon has been pushing the company to allow the use of Claude for “all lawful purposes,” Anthropic’s ethics protocol prohibits its technology from being used to “facilitate violence, develop weapons, or conduct surveillance.”

“It will be an enormous pain in the ass to disentangle, and we are going to make sure they pay a price for forcing our hand like this,” an unnamed Pentagon official told Axios.

Read more

Claude is the only AI model currently deployed on the military’s classified platforms, through a partnership with Palantir Technologies.

The potential blacklisting would require Pentagon contractors to demonstrate that they do not use Anthropic’s technology or risk losing their contracts.

Pentagon spokesman Sean Parnell told Axios that the relationship between the department and Anthropic “is being reviewed.”

“All Pentagon partners must be willing to help our warfighters win in any fight,” Parnell added. An Anthropic spokesperson told the outlet that the company has been having “productive conversations” with the department.

The reported clash follows allegations that the company’s AI model was used during the operation to abduct Venezuelan President Nicolas Maduro in early January.

Last week, Axios and the Wall Street Journal reported that Claude was used both in staging the raid and carrying it out, although its exact role remains unclear. The allegations surfaced after the company spent weeks publicly touting its safeguard policies and presenting itself as the safety-conscious option within the AI sector.