OpenAI Backtracks: Pentagon Banned from Mass Surveillance

On March 3, 2026, Sam Altman attempted to put out a reputational fire. OpenAI made urgent amendments to its recent contract with the US Department of Defense. The document officially states: the company's AI models will not be used for mass domestic surveillance and will not receive automatic connection to NSA or CIA databases without a separate protocol.

From a technical standpoint, monitoring this ban on the Pentagon's air-gapped servers is practically impossible. OpenAI transfers the license and deploys the model weights inside a closed perimeter where the provider has no access to telemetry or logs. These "red lines" are a pure PR buffer. The goal is to calm civilian investors and corporate B2B clients spooked by the rapid militarization of the planet's main AI startup. OpenAI will be unable to track exactly what prompts US Cyber Command analysts are entering, no matter how much it wants to.

Source: Reuters / OpenAI Blog