§ feed · storyline
OpenAI says GPT-5 models were accidentally trained on ...
OpenAI says GPT-5 models were accidentally trained on certain content, according to a statement from the company.
A Perplexity page summarizing OpenAI’s statement that GPT-5 models were accidentally trained on certain content, alongside related discussion.
(The page content shown in results focuses on community commentary.)
§ sources1 publication · timeline below