OpenAI plans to train models on Daybreak Blue session data previously kept off limits
🤖 AI-generated content — The title and summary were produced automatically by artificial intelligence, without human editorial review.
From 29 November, OpenAI will train on Daybreak Blue session data after telling Trusted Access members it had been excluded even when users turned off “improve the model for everyone”; backlash brought a second email promising a separate opt-out. Safety-reviewed interactions can still be used for safety. Researchers warn of a repeat of mathematicians Tristan Buckmaster and Levent Alpöge’s Codex dispute over Navier-Stokes drafts.
Continue in the app — vote & join in ➔Source: The Decoder (AI) · via ahirlevel.hu