Watermark for LLM-Generated Text
ID: f4e329a6-c47e-5525-af96-f241a8dd8c4a
STIX ID: report--f4e329a6-c47e-5525-af96-f241a8dd8c4a
Feed Name: Schneier on Security
Researchers at Google describe a cryptographic watermarking method for LLM-generated text that subtly biases token selection using a secret key, allowing keyed detection and showing promising robustness to post-editing, with reliable detection reportedly achievable on samples as small as ~200 tokens.
Your team is not currently subscribed to this feed. You must subscribe to it in order to see this post.