Hacker News new | past | comments | ask | show | jobs | submit
sureMan6 | on: How Claude marks AI-generated content
Maybe it has no false positive rate
loading story #49256546
loading story #49258152
loading story #49257596
embedding-shape | parent | next
Read said section yourself perhaps.
suddenlybananas | parent | next
That's essentially impossible, unless you mean they didn't measure a false positive rate.
Filligree | root | parent
For watermarked long-form text, it is actually possible. Makes the watermark more fragile, but the math is considerably more forgiving than usual.
loading story #49255906
loading story #49257509
loading story #49258123