Military Worries Simmer at OpenAI, Google As Anthropic Hits Stalemate With Pentagon

Anthropic’s refusal to cooperate with the Defense Department is having ripple effects across Silicon Valley.

The conflict between Anthropic and the Department of Defense over how the Pentagon can use the startup’s AI has turned into a standoff.

The agency has given the company a Friday evening deadline to give it unfettered access to its technology—or else be cut off from working with military contractors. Anthropic CEO Dario Amodei continues to refuse to do so. “Threats do not change our position,” he wrote in a blog post published on Thursday. “We cannot in good conscience accede to their request.”

If the dispute involved only the Pentagon and Anthropic, the feud would still represent an intense, intrigue-filled drama. But what’s now becoming clear is that the feud has implications extending beyond Anthropic, raising questions about what will happen next at Google, OpenAI and the other AI companies that have lined up military deals.

Amodei has refused to budge on two issues: He doesn’t want Anthropic’s AI used to surveil Americans or to make decisions on lethal autonomous weapons without human supervision. And now OpenAI has told the Pentagon that it shares the same concerns, an OpenAI spokesperson said on Friday, declining to comment on how its own negotiations with the Pentagon were proceeding. On Thursday evening, OpenAI CEO Sam Altman told his company staff that he hoped to broker a deal with the Pentagon that would secure those safeguards for OpenAI and could be used industry wide.