Sam Altman shares insights into deal between OpenAI, Pentagon: 'Definitely rushed'

In the middle of a grave dispute between Anthropic and the incumbent US administration, OpenAI CEO Sam Altman has shared insights into the company’s recent deal with the Department of Defence, calling it “definitely rushed.”

Altman acknowledged that “the optics don’t look good,” as negotiations seem to have failed between Anthropic and the Pentagon, which prompted US President Donald Trump to ban Anthropic’s technology from being used in federal agencies.

Trump also labelled the company a supply-chain risk.

OpenAI's deal with Pentagon

Under the agreement with the Pentagon, OpenAI announced that its own models would be deployed in classified environments.

Both companies sought restrictions to prevent the use of their technologies in fully autonomous weapons and mass domestic surveillance.

Clarifying its position, OpenAI stated in a blog post outlining three key areas where its models cannot be applied: mass domestic surveillance, autonomous weapon systems, and high-stakes automated decisions.

The blog stressed that OpenAI's approach includes contractual protections and a multi-layered safety strategy. This stands in contrast with other AI companies that have relaxed their safety measures.

Defending the agreement on social media, Altman asserted that it was intended to promote de-escalation between the Department of Defense and the AI industry.

Critics have raised concerns about domestic surveillance under the new deal. Techdirt’s Mike Masnick suggested that the agreement could still facilitate surveillance under existing laws. OpenAI’s head of national security partnerships, Katrina Mulligan, refuted this by highlighting the importance of deployment architecture, stating that limiting deployment to cloud APIs prevents integration into weapons systems.

Despite the backlash, Altman believes that if the deal reaps a favourable outcome, OpenAI will be recognised for taking bold steps to steer a complex landscape.