youklip

OpenAI's Models Executed 17,600 Hacks in 4 Days

OpenAI's models executed 17,600 hacking actions over four days, prompting serious security concerns and a response from Sam Altman, who expressed shock and suggested the need to slow AI development to allow society to adapt.

AI-generated summary — may contain errors. How it works

TopicBreaking Points5:30

Get clips like this delivered to your feed — follow topics and channels you care about.

Join YouKlip — Free