Every time you open ChatGPT, there’s an assumption baked into the interaction you probably don’t notice: that you’re talking to a company. A private one. Something with a Terms of Service, a support email, maybe an IPO on the horizon — but fundamentally a business, separate from the state, answerable to shareholders and regulators the way any SaaS product is.
That assumption just got a lot shakier.
The Pivot
OpenAI has reportedly floated giving the U.S. government a 5% equity stake in the company — worth an estimated $42.6 billion at OpenAI’s current $852 billion valuation — as part of a broader plan to have every leading American AI lab contribute a similar share to a public investment vehicle modeled directly on the Alaska Permanent Fund, the sovereign fund that turns Alaska’s oil revenue into annual resident dividends.
This isn’t a stray comment. CEO Sam Altman has personally raised the idea in conversations with President Trump, Commerce Secretary Howard Lutnick, and Treasury Secretary Scott Bessent — and has separately discussed it with Senator Bernie Sanders, who’s pushing an even more aggressive version: a 50% public ownership stake across major AI companies, funded through a one-time equity tax, that he estimates could be worth $7 trillion and fund $1,000 annual dividends for every American.
To be clear about where this stands: the talks are described as conceptual and early-stage, and any real deal would likely need an act of Congress. Nothing is signed. But the direction of travel is already visible in the government’s other moves — Washington took a 9.9% stake in Intel last year by converting CHIPS Act funding into equity, and it’s reportedly holding stakes or “golden shares” in roughly twenty private companies already, spanning minerals, semiconductors, and quantum computing.
What makes the AI version different is what the product actually does. Intel makes chips. The government owning a piece of Intel doesn’t change what happens when you plug in a laptop. But if Washington ends up holding equity in OpenAI, the government has a financial and structural relationship with the same system you might ask about a medical symptom, a legal question, a divorce, a mental health spiral, or a message you’re not sure how to send to your boss. The tool that reads your most unguarded questions would have a shareholder that also happens to write export control law, subpoena tech companies, and regulate the exact industry the shareholder is invested in.
That’s not a normal customer-company relationship anymore. It’s something closer to a utility with a badge.
So What?
We already made peace with surveillance-adjacent technology once — we did it with search engines, social platforms, telecoms — by telling ourselves they were “just companies,” bound by contracts and courts like anyone else, with government as an occasional outside referee rather than a co-owner. That framing did a lot of quiet work. It let us type things into a text box we’d never say out loud, on the theory that the entity on the other end had commercial incentives, not state ones.
If the user agreement and the national interest start to become the same document, that theory stops holding. The question isn’t necessarily whether this specific deal goes through — reporting suggests it may not, and rivals like Anthropic haven’t signaled they’d join in. The question is what you do with the fact that it’s now a live, serious proposal at all, discussed at the level of the Treasury Secretary rather than dismissed as fringe.
So: next time you’re about to type something private into an AI chatbot, it’s worth asking a question that used to sound paranoid and now just sounds current — whose interests, exactly, sit on the other side of this conversation?