Third-party cyber evaluations involving OpenAI models

(openai.com)

36 points | by glub 3 hours ago ago

4 comments

  • cadamsdotcom 31 minutes ago

    Any testing of cyber capability in a sandbox should be prefaced with a test where the model is tasked with escaping the sandbox ;)

    Smoke out those misconfigurations while the model only needs to escape, not do anything once out.

  • solenoid0937 2 hours ago

    Wait, is Irregular the same company that caused the Anthropic incident?

    • dnw 2 hours ago

      yes

  • wmf 21 minutes ago

    Now that is a vague headline. Is the secret ingredient crime?