Anthropic Says Pentagon Talks on Claude Safeguards Hit a Wall
CEO Dario Amodei admits contract negotiations with the DOD have made almost zero headway on critical AI safety guardrails.
Anthropic's negotiations with the Department of Defense over how Claude can be used aren't going well. At all.
CEO Dario Amodei said Thursday that new contract language proposed by the Pentagon has made "virtually no progress" on two key concerns: preventing Claude from being deployed for mass domestic surveillance and blocking its use in fully autonomous weapons systems.
The stalemate puts a spotlight on the growing tension between AI companies and military adoption. Anthropic has positioned itself as the safety-first AI lab, but that stance is now being tested against the realities of government contracting.
The DOD apparently wants access to Claude's capabilities without the restrictions Anthropic considers non-negotiable. Whether either side blinks remains to be seen — but right now, the gap looks wide.