4 ms·

Interesting write-up and I do think LLM assisted/powered exploit disclosure is a real concern (I've been able to get models to create container breakouts from L

Interesting write-up and I do think LLM assisted/powered exploit disclosure is a real concern (I've been able to get models to create container breakouts from Linux LPEs relatively quickly). One thing I'm surprised about is that GPT-5.6 didn't block that prompt due to guardrails. My experience is that GPT-5.5 and up does not like offensive security work (similar to Opus 4.7+/Fable). I didn't notice it but I'd assume that the authors have some level of cyber approvals from OpenAI to relax the guardrails a bit.

Santas 2mo ago
This might help https://chatgpt.com/cyber https://chatgpt.com/cyber ease the guardrails a bit.
- eru 2mo ago
  Thanks! I wonder if Claude has something similar?
  - NiekvdMaas 2mo ago
    They do: https://portal.anthropic.com/programs/cvp https://portal.anthropic.com/programs/cvp
    - danslo 2mo ago
      Though this program does not apply to Fable. Which is why most security researchers have started flocking to Sol.