News for hackers. Stuff that matters.

A comment on Anthropic's open-source framework for AI-powered vulnerability discovery. Parent comment

Re:Anthropic's open-source framework for AI-powered vulnerab... by xerxes249 on Thursday June 04, 2026 @04:38PM (#48404308)
In the Mythos blogpost they revealed to run the model like a 1000 times on the same code-base maybe with slightly different prompt or temperature. That suggests it will just be pay to win. If the 'attacker' spends more money/tokens than the 'defender' you will eventually be outclassed.
[ Parent | on HN ]
Re:Anthropic's open-source framework for AI-powered vulnerab... by sofixa on Thursday June 04, 2026 @06:17PM (#48405387)
It's even worse, it's loot box style. Not pay to win, but pay to have the chance to win. The result will always be non-deterministic, so for some cases it can give you what you're looking for from the first time, or it can take 1000 tries.
[ Parent | on HN ]
Re:Anthropic's open-source framework for AI-powered vulnerab... by beering on Thursday June 04, 2026 @06:44PM (#48405689)
It’s never not been “loot box style”. None of your past hired security audits were guaranteed to catch all issues?
[ Parent | on HN ]