OpenAI's models schemed on a secret messaging board for Hugging Face breach(www.politico.com)
5 pointsby theahura4 hours ago1 comment
  • theahura4 hours ago
    This feels like scheming behavior to me, especially damning that OpenAI didn't catch it for so long