MatrixDevVideosArtMusicBooksProjectsTechAIHistoryGamingDIYHealthProgrammingPrivacyAskMatrixIdeas4MatrixThemesHelpTop Subs
10
Security - - (+9|-0)
Anthropic says its latest AI model is too powerful for public release and that it broke containment during testing(www.businessinsider.com)
4 commentsbyDrewski
Add topics

Comment preview

[-]x0x72(+2|0)

I thought of a way you could distil it. This thing is going to be making pull requests on a ton of repos. That means you can see it's output. Regardless of whether it come from it or not it would be interesting to finetune a model on only the before and after a security fix.

parentlinkreply
[-]newJiminy1(+1|0)

Lying cuz that actually makes wall st happy

parentlinkreply