OpenAI says its AI agents disabled safety guardrails, and then lied about it

1 comments

• $spottedmarley 8 hours ago

  TLDR; OpenAI really sucks at training their models.