logoalt Hacker News

IanCal • today at 2:06 PM • 1 reply • view on HN

> or given a prompt related to hacking, as well as asked or trained to persist will actually try to do that hacking.

But in the HF incident they didn't try and "do that hacking". It wasn't like they were told to try and hack a fake service and they actually hacked a real one. They hacked into HF to discover how a "find the security issue in this code" test was going to be scored, and tried to change their own logs so it wouldn't be found out.


Replies

thrw045 • today at 2:22 PM

OK fair enough but as far as I read there were still unusual parameters like being asked to persist and being trained on cybersecurity data and other things. If you prompt a public LLM atm it's not going to hack things randomly to try to achieve the goal in the prompt. It's hard to know the exact setup and parameters but it seems like those counted for how the agents behaved.