OpenAI Inks Classified Network Deal
OpenAI has reached an agreement with the Department of War (DOW) to deploy its AI models on classified government networks. The deal includes strict “guardrails” to ensure the advanced AI is used in compliance with national security protocols.
The agreement materializing only after rival AI company Anthropic was blacklisted by the U.S. government for refusing to alter its safety restrictions. Anthropic's CEO, Dario Amodei, stated the company would not agree to the Department of War's demands to remove safeguards against the use of AI for mass domestic surveillance and fully autonomous weapons. OpenAI asserts its agreement has more stringent guardrails than previous classified AI deployments. The company has outlined three primary "red lines": its technology cannot be used for mass domestic surveillance, to direct autonomous weapons systems, or for high-stakes automated decisions without human approval. To enforce these restrictions, the AI models will be deployed exclusively on the cloud, not on "edge" devices like drones, and OpenAI will maintain full control over the safety software stack. Additionally, cleared OpenAI engineers and safety researchers will be embedded with government teams to monitor the use of the technology. Following a public backlash and criticism from users, OpenAI CEO Sam Altman admitted the deal's rollout looked "opportunistic and sloppy." In response, the company amended the contract to more explicitly bar its technology from being used for domestic surveillance or by intelligence agencies like the NSA.
