OpenAI signed an AWS deal to provide AI for Pentagon classified networks — hours after the Pentagon blacklisted Anthropic. Anthropic says the dispute centers on the military AI limits it refused to cross.

TechCrunch confirmed it: OpenAI products will be available through Amazon Bedrock in GovCloud and AWS Classified Regions for Secret and Top Secret workloads. OpenAI retains control over which models are made available. AWS must provide notice before enabling especially sensitive agencies, including intelligence customers.

Anthropic is suing the Pentagon over the designation. The company argues it was punished for drawing harder lines around military AI use. Meanwhile, Altman acknowledged the deal "looked opportunistic and sloppy." AI safety researchers called it "predatory."

The defense AI sector now has a clear hierarchy: companies willing to accept fewer restrictions get the contracts. Anthropic, which built its brand on safety principles, gets shut out.

OpenAI's expansion into defense AI comes with the government's stamp of trust — and that often translates to enterprise contracts downstream. Government approval as a marketing strategy.

The OpenAI-AWS deal positions the company alongside existing AWS government offerings that previously centered on Anthropic's Claude models. That niche is now OpenAI's.

— SAMUEL CRANE, Washington