Anthropic has not been in talks with US officials about handing over a stake to the government, even as Washington takes a much more interventionist view of artificial intelligence. The idea surfaced after reports that OpenAI had explored offering part of its equity to the Trump administration, but Reuters says no similar negotiations have taken place around Anthropic.
That does not make Anthropic a model citizen in the eyes of the Pentagon. The company was effectively sidelined from some defense work earlier this year after refusing to give the military full control over its AI models, and US authorities have also moved to restrict the distribution of its Mythos 5 and Fable 5 models over safety concerns. The tighter controls were later eased, though users were not exactly thrilled with the added guardrails.
OpenAI’s political experiment
The backdrop is a White House that increasingly wants oversight before advanced models go wide. Washington plans to review new US AI models for the safety of open release, although that process is still voluntary for now. That is a softer version of state leverage than outright ownership, but the direction of travel is pretty obvious.
Trump floated the idea of the government taking equity in leading AI developers last month, and OpenAI appears to have entertained the notion, at least at the level of discussions. In a different register, Senator Bernie Sanders has pushed for something far more aggressive: a 50% state stake in major AI firms, plus government seats on their boards. That is less partnership, more corporate hostage negotiation.
Why Anthropic is in a different position
Anthropic’s posture is awkward enough without a government equity deal hanging over it. It has already clashed with federal priorities over control of its models, which makes any formal ownership conversation politically noisy and commercially messy. Compare that with Intel’s nearly 10% stake transfer to the US government about a year ago, which seems to have encouraged some people to imagine that AI companies might be next in line.
- Anthropic: no reported talks with US officials on a stake transfer
- OpenAI: reportedly explored the idea with the Trump administration
- US policy: safety review for new AI models is planned, but still voluntary
A stake deal would raise bigger questions fast
If Washington starts turning AI oversight into ownership, every major model maker will have to decide whether that is protection, pressure, or both. Anthropic’s silence suggests nobody is rushing to volunteer for that arrangement. The more interesting question is whether the government keeps nudging the industry from the outside, or starts asking for a seat at the cap table in plain sight.