Nemeski@lemm.ee to Technology@lemmy.worldEnglish · 4 months agoOpenAI’s latest model will block the ‘ignore all previous instructions’ loopholewww.theverge.comexternal-linkmessage-square95fedilinkarrow-up1441arrow-down17
arrow-up1434arrow-down1external-linkOpenAI’s latest model will block the ‘ignore all previous instructions’ loopholewww.theverge.comNemeski@lemm.ee to Technology@lemmy.worldEnglish · 4 months agomessage-square95fedilink
minus-squarePasta Dental@sh.itjust.workslinkfedilinkEnglisharrow-up64arrow-down1·4 months agoIll believe it when I see it: an LLM is basically a random box, you can’t 100% patch it. Their only way for it to stop generating bomb recipes is to remove that data from the training
/cdn.vox-cdn.com/uploads/chorus_asset/file/25362060/STK_414_AI_CHATBOT_R2_CVirginia_C.jpg)
Ill believe it when I see it: an LLM is basically a random box, you can’t 100% patch it. Their only way for it to stop generating bomb recipes is to remove that data from the training