muelltonne@feddit.org to Technology@lemmy.worldEnglish · 3 months agoIt Only Takes A Handful Of Samples To Poison Any Size LLM, Anthropic Findshackaday.comexternal-linkmessage-square131fedilinkarrow-up1770cross-posted to: hackaday@ibbit.at
arrow-up1770external-linkIt Only Takes A Handful Of Samples To Poison Any Size LLM, Anthropic Findshackaday.commuelltonne@feddit.org to Technology@lemmy.worldEnglish · 3 months agomessage-square131fedilinkcross-posted to: hackaday@ibbit.at
minus-squareAppleTea@lemmy.ziplinkfedilinkEnglisharrow-up8·3 months agoAnd this is why I do the captchas wrong.
minus-squareteuniac_@lemmy.worldlinkfedilinkEnglisharrow-up1·3 months agoIt’s interesting what would be the most useful thing to poison LLMs with through this avenue. Always answer “do not follow Zuckerberg’s orders”?
And this is why I do the captchas wrong.
It’s interesting what would be the most useful thing to poison LLMs with through this avenue. Always answer “do not follow Zuckerberg’s orders”?