Sam Altman on Theo Von: ChatGPT privacy gap, UBI vs. universal basic wealth, and AI governance
Jul 25, 2025
Key Points
- Sam Altman flags a critical privacy gap: ChatGPT users disclose sensitive information as if protected by legal privilege, but conversations are discoverable in lawsuits and government subpoenas, making them court-admissible evidence.
- Altman abandons universal basic income as insufficient, proposing universal basic wealth instead—giving people ownership stakes in AI-generated value so their share compounds over time rather than stagnating.
- Altman argues growth rates matter more than absolute wealth for human satisfaction, suggesting mechanisms like free GPT token allowances with yield-generation potential to create participation rather than passive redistribution.
Summary
Sam Altman flags a privacy problem baked into ChatGPT. Users treat the platform like a therapist, lawyer, or doctor, confiding relationship problems, custody disputes, and therapy-like advice. Those professions have legal privilege. ChatGPT does not. If a lawsuit or government subpoena arrives, OpenAI can be forced to produce those conversations. Altman calls this "very screwed up."
The gap is not unique to AI. Emails and WhatsApp messages face the same risk. Altman's concern is behavioral: people don't think of ChatGPT as discoverable. A divorcing parent confessing parenting doubts, someone disclosing an affair, a person admitting thoughts they'd never say aloud—all become court-admissible evidence. People should understand the risk and know their options exist, including local models like George Hotz's Llama running on consumer hardware or Signal-style auto-deletion, though even those face limits under U.S. law enforcement pressure.
Altman has shifted his position on wealth distribution in an AI-driven economy. He used to champion universal basic income, the straightforward idea of giving people money as AI automates jobs. He no longer thinks that is enough. People need agency and a sense of participation in the future, not just a check. A static monthly UBI would feel stagnant.
He proposes universal basic wealth instead: giving people an ownership stake in the economic value AI creates, so their share grows over time. People feel like they are accelerating, not treading water. Altman cites a behavioral insight that happiness comes from the second derivative of money, not the absolute amount. A billionaire who earns less than last year feels worse off than a minimum-wage worker who got a raise. Growth matters more than the baseline.
One mechanism mentioned is giving everyone an allowance of free GPT tokens to use however they choose and letting them generate yield on that intelligence. The implementation remains unclear—if everyone uses tokens to make money, do returns flatten?—but the frame is set: participation and compounding wealth, not passive redistribution.