Anthropic CEO accuses OpenAI of ‘straight up lies’ over Pentagon deal

The chief executive of Anthropic has sharply criticised rival OpenAI over its recent defence contract with the United States government, accusing the company of misrepresenting the circumstances surrounding the agreement.

In a memo to staff reported by The Information, Anthropic’s co-founder and CEO, Dario Amodei, described OpenAI’s handling of its deal with the US Department of Defense as “safety theatre”. He wrote that OpenAI had prioritised placating employees, while Anthropic had focused on preventing potential abuses of its technology.

The dispute follows failed negotiations between Anthropic and the US Department of Defense over a new contract that would have expanded the military’s access to the company’s AI systems. Anthropic, which previously held a $200 million defence contract, sought explicit assurances that its models would not be used for domestic mass surveillance or fully autonomous weapons. The Department instead reached an agreement with OpenAI.

In a recent blog post, OpenAI stated that its systems could be used for “all lawful purposes” and that the Department had affirmed that mass domestic surveillance was illegal and not planned. The company added that this understanding had been made explicit in its contract.

In his internal message, Amodei referred to OpenAI’s public statements as “straight up lies”, accusing Altman of presenting himself as a mediator while downplaying substantive differences between the companies’ positions. Anthropic’s concerns reportedly centred on the Pentagon’s insistence that its AI tools be available for any lawful use, a formulation that could shift if legal standards change over time.

Following news of OpenAI’s defence deal, downloads of its ChatGPT application reportedly fell sharply, while Anthropic’s app rankings improved.

Read more

Users are mass-uninstalling ChatGPT after Pentagon deal, data shows