Anthropic Mythos model can find and exploit 0-days
www.theregister.com
: Hasn't released it to the public, because it would break the internet - in a bad way

Hasn’t released it to the public, because it would break the internet - in a bad way For years, the infosec community’s biggest existential worry has been quantum computers blowing away all classical encryption and revealing the world’s secrets. Now they have a new Big Bad: an AI model that can generate zero-day vulnerabilities.…

  • ℍ𝕂-𝟞𝟝@sopuli.xyzEnglish
    3·
    28 days ago

    I actually have a model that is 10 percentage points better than Anthropic’s on my old laptop, it actually found ten zero days exploitable within Claude itself.

    In the interest of humanity, I’m not letting anyone see it. /s