Anthropic CEO Blasts OpenAI’s Pentagon Deal Messaging as ‘Straight Up Lies’

AI safety tensions escalate as competing companies clash over military contracts and ethical boundaries

Anthropic CEO Dario Amodei has launched a scathing attack on OpenAI’s public messaging around its recent Pentagon contract, calling the company’s statements “straight up lies” in what marks one of the most direct confrontations yet between leading AI companies over military partnerships.

The Contract Handoff

The controversy stems from Anthropic’s decision to walk away from a lucrative Pentagon contract earlier this year due to fundamental disagreements over AI safety protocols and military applications. OpenAI quickly moved to fill the vacuum, securing the contract and positioning itself as a more collaborative partner for defense applications.

Amodei’s explosive comments, first reported by industry insiders, suggest deep frustration with how OpenAI has characterized both the transition and its own commitment to responsible AI development in military contexts.

Safety First vs. Strategic Partnership

Anthropic, which has built its brand around AI safety and responsible development, reportedly terminated its Pentagon discussions when military officials pushed for capabilities that the company deemed potentially dangerous or ethically problematic. Sources familiar with the negotiations indicate the disagreements centered on autonomous decision-making systems and the extent of human oversight required.

“We’ve always been clear that there are certain lines we won’t cross, even if it means walking away from significant revenue,” said one Anthropic executive, speaking on condition of anonymity.

OpenAI’s approach appears markedly different. The company has framed its Pentagon partnership as a way to ensure American military superiority while maintaining ethical guardrails—a position that Amodei apparently views as contradictory or misleading.

The Broader Industry Divide

Th