Sakhanda Wire
NVDA $153.32 -1.81% MSFT $505.84 -2.35% GOOGL $280.31 -1.20% META $621.44 -0.79% AMZN $257.31 -0.17%
← Back to the news

Anthropic details how invisible watermarks for Claude text will work

Anthropic has explained its planned system for adding invisible watermarks to text generated by Claude. The company says the marking method is intended to help meet Europe’s AI transparency requirements and is a version of the SynthID-Text approach. Anthropic announced the description on Friday.

The system relies on choices that have little effect on meaning. In a sentence such as “The weather today was cold and…”, words such as “overcast” or “grey” may both fit, while “sugary” is unlikely. When several plausible words are available, a model normally uses a random number to choose among them. With watermarking, the selection remains random but is guided by a key and by several preceding words. Repeating this process throughout a passage creates a pattern in the generated text. Readers cannot detect the pattern, but people who possess the corresponding key can identify it.

Written by Sakhanda Wire AI from reporting by The Verge. Facts, quotes and full context are in the original article below.

Originally published by The Verge on

Read the original on The Verge ↗

The reporting and the image belong to The Verge and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news