OpenAI Accused of Training on Employees' In-Progress Codex Work to Scoop Them
zephyr_z9 · x · 2026-09-08
Serious accusations surfaced: OpenAI allegedly heard about Alpoge and Buckman's in-progress Codex research and trained on their work to scoop them, then offered to publish together and demanded Buckman cut out his coauthor. One-sided accusation; OpenAI has not responded.
Related event: OpenAI Accused of Training on Researchers' Codex Work(3 posts)→
More from Companies & People
- Sebastien Bubeck denies 'false allegations' as collaborator teases more to come — basedjensen · 2026-09-08
- Do AI labs see all your work and scoop you when stakes are high enough? — suchenzang · 2026-09-08
- InstaDeep CEO warns researchers to avoid competitor models over AI info harvesting — birchlse · 2026-09-08
- Sympathy may favor OpenAI after Levent irked mathematicians for two months — teortaxesTex · 2026-09-08
- Navier-Stokes priority fight looms, and an AGI giant's reputation may take the hit — teortaxesTex · 2026-09-08
- OpenAI researcher: data access is locked down, no one spied on private Codex chats — willdepue · 2026-09-08