Stenberg: Mythos finds a curl vulnerability

Daniel Stenberg has published a lengthyarticle on his thoughts on Anthropic's Mythos, which the companydecided was too dangerous for wide public release.

My personal conclusion can however not end up with anything elsethan that the big hype around this model so far was primarilymarketing. I see no evidence that this setup finds issues to anyparticular higher or more advanced degree than the other tools havedone before Mythos. Maybe this model is a little bit better, but evenif it is, it is not better to a degree that seems to make asignificant dent in code analyzing.

This is just one source code repository and maybe it is much betteron other things. I can only tell and comment on what it foundhere.

But allow me to highlight and reiterate what I have said before: AIpowered code analyzers are significantly better at finding securityflaws and mistakes in source code than any traditional code analyzersdid in the past. All modern AI models are good at this now. Anyonewith time and some experimental spirits can find security problemsnow. The highquality chaos is real.