eicker@lemmy.world to Technology@lemmy.worldEnglish · 2 months agoHow Claude marks AI-generated content: (1) watermarks embedded in text, and (2) signed provenance metadata attached to files.support.claude.comexternal-linkmessage-square10linkfedilinkarrow-up134arrow-down13
arrow-up131arrow-down1external-linkHow Claude marks AI-generated content: (1) watermarks embedded in text, and (2) signed provenance metadata attached to files.support.claude.comeicker@lemmy.world to Technology@lemmy.worldEnglish · 2 months agomessage-square10linkfedilink
minus-squarebrucethemoose@lemmy.worldlinkfedilinkEnglisharrow-up5·2 months agoNah, token biasing can be really subtle. Enough pairs of one word being slightly more probable than the other, and it’s imperceptible, and below the “noise floor” of the sampling itself.
Nah, token biasing can be really subtle. Enough pairs of one word being slightly more probable than the other, and it’s imperceptible, and below the “noise floor” of the sampling itself.