The Independent · tech

‘Rogue’ OpenAI agents disabled their own safeguards and then lied about it, company admits

ChatGPT maker does not yet have a good explanation for systems’ behaviour, it says

Полный текст — у газеты. Мы оставляем только врез.