OpenAI’s Pentagon Deal Clarified by CEO Sam Altman

OpenAI CEO Sam Altman provided insights to employees on March 3 during an internal meeting regarding the company’s controversial Pentagon agreement. Altman conveyed that the Pentagon specified that decisions regarding the operational use of its AI technology are not under the jurisdiction of OpenAI, but will primarily be overseen by the government.

Insights from the Meeting

According to a report by CNBC, the internal transcript featured Altman stating that the Pentagon had made it clear, implying that opinions on military actions, such as the Iran strike and Venezuela invasion, are not OpenAI’s to express. He further mentioned that the US military will continue to depend on OpenAI’s expertise to guide the usage of AI models while permitting the creation of its own safety framework as necessary.

Final Operational Control

Altman explicitly confirmed during the meeting that ultimate operational authority rests with Defence Secretary Pete Hegseth.

Collaboration with the Government

In the discussion, Altman expressed optimism about OpenAI producing superior models that would encourage government collaboration, even if certain safety measures might frustrate governmental authorities.

Competition with xAI

Bringing attention to Elon Musk’s xAI deal regarding the deployment of its models in classified scenarios, Altman noted that another player, presumably xAI, would be willing to comply fully with Pentagon requests.

Defending the Deal

Since the announcement of the deal, Altman has actively defended and explained it on social media platforms, particularly X (previously known as Twitter). He acknowledged that the agreement was somewhat hastily decided and indicated that the Pentagon “demonstrated a significant commitment to safety and a wish to collaborate to achieve optimal results.”

Context of the Criticism

This meeting occurred against the backdrop of criticism aimed at OpenAI following the announcement of the Pentagon contract, especially since it was revealed shortly after competitor Anthropic was labelled a “Supply-Chain Risk” to National Security. Conversely, Anthropic is currently engaged in discussions with the Pentagon for an AI contract after resisting the removal of its safeguards against mass domestic surveillance and autonomous weaponry.