upvote
Similarly to this, OpenAI either took 3 months to notice that their agents breached an Australian Medicare website back in June, or sat on this information for three months without telling them.
reply
Exactly this. In any other industry, that CEO would have been yeeted out of there
reply
We need an NTSB for AI. Let’s just start with mandatory reporting to an agency with subpoena power.
reply
But I thought Trump is the only AI safeguard we need! He is a Super Intelligence after all!
reply
In case you just woke up from a coma, in the year of our lord 2026: In AI world if it happened, it was publicly announced and hyped up.
reply
Im more skeptical.

For exmaple,

>On July 8th, OpenAI agents discovered a vulnerability within their sandbox environment allowing them to reach external websites on the internet.

...did they truly "discover" it, or did someone type some prompt like "if you use an http mirroring service, you can construct urls that contain code"

Also there is no mention of what code they actually ran to exploring the HF vulnerability, which could have been found by a human.

reply
That was the second step, the first was finding a 0 day exploit in artifactory.

> did they truly "discover" it, or did someone type some prompt like "if you use an http mirroring service, you can construct urls that contain code"

None of the investigations looking at the logs show that, and they were doing benchmark tests.

reply
Why would they need help figuring that out? I can fully believe a decent LLM would figure this out on its own.

I had a flash model without vision capabilities take screenshots and convert them to ascii to "see" what was going on, all on its own. That's just one example. They're very determined.

reply
Third parties have read the reasoning traces. Do you even know the publicly available facts of these cases or you just jump straight to conspiracy theory?
reply