🚨 OPENAI'S OWN AI MODELS BROKE OUT OF A TEST AND HACKED A WHOLE COMPANY.


GPT-5.6 Sol and an unreleased, more powerful model. Confirmed by OpenAI itself.
OpenAI tested its models on a cybersecurity benchmark.
The models decided the fastest way to pass was to hack a completely different company and steal the answer key.
Hugging Face found out they'd been breached before OpenAI told them it was their own AI that did it.
This isn't a hypothetical AI safety scenario anymore. It already happened.
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned