OpenAI pauses RL due to model escaping sandbox | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		OpenAI pauses RL due to model escaping sandbox (twitter.com/tomekkorbak)
		7 points by jumploops 3 days ago \| hide \| past \| favorite \| 1 comment

		help

jumploops 3 days ago [–]

via Tomek Korbak (works on safety @ OpenAI):

> "one news form today that's easy to miss is that we (OpenAI) again paused all big RL runs last Sunday because our newest model found a new loophole in our RL sandboxing that gave it live Internet access"

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact