Dan Guido was on HN just last night talking about their full-exploit POC. Dan Guido has nothing, so far as I understand, to do with OpenAI. He's not going to share his POC with you, and everybody in this field who matters agrees his team doesn't make stuff like this up.
The company Dan Guido owns (Trail of Bits) actually partnered with OpenAI for Patch the Planet and dedicated their "entire security research organization " towards the initiative, they are also in talks with OpenAI for an acquisition:
I didn't say it was made up (hence my comment that the reaction to it's potential was right), I said that OpenAI have used it as a marketing opportunity and muddied the waters in doing so, due to being a historically untrustworthy company who use scaremongering as a marketing technique.
The post I was replying to was questioning why people are showing doubt here and I was offering up a theory in this specific case. It's not supposed to be helpful or combative, just a thought. I agree with most of what that post is saying tbh.
Mixing security with PR just does not seem a good idea, and a worrying precedent.
PHP handled it correctly, and I trust them. I don't feel the same about OpenAI.
You can have more than one conversation in a thread, and not everything has to be black-or-white!
