Comment by Analemma_
17 days ago
“Opus 4.6 found 22 security bugs, Mythos found 271 on an initial evaluation” sure seems to refute the grumbling I’ve seen from a couple OAI people on Twitter that Mythos isn’t actually anything special and everything it finds could be found by earlier models too.
They also put this in the end in boldfaced:
"Encouragingly, we also haven’t seen any bugs that couldn’t have been found by an elite human researcher."
But, in overall, I think it was a well-written positive take (instead of the fear-mongering party line).