OpenAI shares more details about its agreement with the Pentagon | TechCrunch

"We retain full discretion over our safety stack, we deploy via cloud, cleared OpenAI personnel are in the loop, and we have strong contractual protections. This is all in addition to the strong existing protections in U.S. law."

"In contrast to other AI companies that have reduced or removed their safety guardrails and relied primarily on usage policies as their primary safeguards in national security deployments, OpenAI's agreement protects its red lines through a more expansive, multi-layered approach."

"With Anthropic saying it was drawing red lines around the use of its technology in fully autonomous weapons or mass domestic surveillance, and Altman saying OpenAI had the same red lines, there were some obvious questions: Was OpenAI being honest about its safeguards?"

OpenAI announced a deal with the Department of Defense to deploy AI models in classified environments following failed negotiations between Anthropic and the Pentagon. President Trump directed federal agencies to cease using Anthropic's technology after a six-month transition period, with Defense Secretary Pete Hegseth designating Anthropic as a supply-chain risk. Both companies stated they maintain red lines against autonomous weapons and mass surveillance, yet OpenAI successfully reached an agreement while Anthropic did not. OpenAI published a blog post detailing its approach, emphasizing three prohibited use cases: mass domestic surveillance, autonomous weapon systems, and high-stakes automated decisions. The company claims its multi-layered safeguards—including retained safety stack discretion, cloud deployment, cleared personnel involvement, and contractual protections—distinguish it from competitors that reduced safety guardrails.

Read at TechCrunch