[ianm218]

I think your narratives are a bit mixed up. We went from science experiment ChatGpt 2022 to zero day finder that is too dangerous to release 4 years later. So whoever has been talking about progressive model updates has surely been correct?

[sublinear]

Plenty of zero days have also been found with brute force and random fuzzing.

I'm not downplaying the LLMs we have now, but none of this got started in 2022 apart from the hype train.