Timeline Doesn't Add Up: Writer Dismisses OpenAI's Data Theft Claims Against DeepSeek
Dan_Jeffries1 · x · 2026-07-23
Author Dan Jeffries pushes back against OpenAI's accusations that DeepSeek stole its model data, dismantling the claims through a timeline analysis.
He points out that OpenAI's Fable model was publicly available on June 9, taken down three days later, and unavailable until June 30. DeepSeek released its model on July 16. He argues this brief window is insufficient for a meaningful data theft at the tail end of a massive training run.
If OpenAI claims their Project Glasswing partners were hacked, that requires solid evidence rather than press releases and tweets. The author emphasizes his desire for American open-source models to win based on merit, not by spinning "fables without evidence."
More from Companies & People
- Post says the U.S. led open models first, then let closed models take over — cocktailpeanut · 2026-07-23
- For enterprise AI agents, the first question is ownership and governance, not Copilot Studio vs. Foundry — WirelessLife · 2026-07-23
- WAIC speaker says world models are gaining real momentum in China — smtabatabaie · 2026-07-23
- User criticizes Anthropic strategy: RLHF unbearable, Chinese models more practical — StewartalsopIII · 2026-07-23
- AI-culture joke says you now have to LARP to succeed — KadriJibraan · 2026-07-23
- NERC robotics conference moves to Princeton with papers, demos, and Rising Stars — tomssilver · 2026-07-23