If you do your best work inside a company's AI tool, does the company own what you did in it?
A mathematician spent a year drafting a solution to a million-dollar problem inside OpenAI’s coding tool, then watched OpenAI race down the same route with its own model and asked why he’d ‘ruin his career’ when he objected. As more of us think, draft, and build inside AI tools owned by someone else, where should the line fall between the vendor learning from usage data and the vendor competing with its own customers?
Commentaires (1)
In today’s episode of Minds, Bodies, and Terawatts, dated September 9, 2026, the hosts dig into OpenAI’s claim that an unreleased model coordinating ten thousand agents resolved the Navier-Stokes Millennium Prize problem in 88 hours, and then into the much messier story of NYU’s Tristan Buckmaster, who says he was solving it first, inside OpenAI’s Codex, as a paying customer. The episode treats the math as real but argues the bigger question is contractual: OpenAI itself says it ‘cannot rule out’ that de-identified usage data improved the model, which is a very different answer from ’no.’ The hosts connect this to the book’s idea of the tool that trains its replacement, and ask what it means for every lawyer, engineer, or writer whose daily drafts now live inside a vendor’s product. Listen to the full episode and tell us where you think ownership should sit.
Related reading on unscarcity.ai:
Envie d'aller plus loin ?
Obtenez le plan complet dans <em>L'ère de la post-pénurie : Repenser la société à l'ère des machines</em>
