You are viewing a single comment's thread from:

RE: OpenAI's new cyber model found 81 Firefox bugs nobody knew about, and almost nobody can use it yet

in LeoFinance • 13 days ago

That 0.07% number is the one I kept coming back to as well. A model that finds flaws is impressive, but one that stays inside its limits is the part that actually makes it usable in real life. The honeypot test was a smart idea after the Hugging Face incident. Good breakdown of the benchmarks too.

Sort:  

Yeah, that honeypot number is the part I kept too. Finding flaws is one thing, staying inside the lines is what makes it safe to actually use.