OpenAI CEO Sam Altman informed employees that the company lacks authority over the Defense Department’s decisions regarding its artificial intelligence software. He indicated this stance may contribute to ongoing tensions between the Pentagon and competitor Anthropic.
Key Remarks in All-Hands Meeting
During Tuesday’s company-wide meeting, Altman emphasized that the Defense Department values OpenAI’s technical expertise on AI applications but expects the firm to refrain from judging the merits of specific military operations. “You do not get to make operational decisions,” Altman stated.
Recent Pentagon Agreement
OpenAI finalized a deal late Friday allowing the Pentagon to integrate its AI models into classified networks. This agreement followed disputes with Anthropic, which insisted its technology avoid applications like mass surveillance of U.S. citizens or fully autonomous weapons.
Anthropic raised concerns about potential uses of its tech in operations, including the raid capturing Venezuelan President Nicolás Maduro, though the company denies engaging in discussions of specific missions with the Defense Department.
Alignment with OpenAI Principles
Altman confirmed the pact upholds OpenAI’s guidelines, prohibiting domestic mass surveillance and mandating human oversight for any use of force, including autonomous systems. He described the initial arrangement as “opportunistic and sloppy” and noted ongoing efforts to strengthen it, ensuring agencies like the National Security Agency cannot depend on OpenAI services for such purposes.
Addressing Anthropic Tensions
Altman expressed intent to urge the Defense Department to revoke its unprecedented designation of Anthropic as a supply-chain risk—a label typically reserved for U.S. adversaries. He aims to help resolve the conflict between the Pentagon and Anthropic.
