Evidence · DiggingBeagle record
Codebook watermarks recover short-text identifiers under selected edits (research paper)
A cooperative language model encodes short identifiers through selected output-token bias, but the extraction rate falls under some editing and instruction-tuned variants.
- Published
- Nov 12, 2026
- Source role
- primary disclosure
Evidence record
A cooperative language model encodes short identifiers through selected output-token bias, but the extraction rate falls under some editing and instruction-tuned variants.
Claim-level citations (4)
- supportsCodebook watermarks recover short-text identifiers under selected edits: The paper describes nearest-codeword identification as a search over the keyed codebook. The number of candidate codewords grows exponentially with identifier bit length, so short-identifier retrieval results do not by themselves establish affordable decoding for a much larger identity space.
Section VI, discussion of codebook size, nearest-codeword decoding and scaling limitations
- supportsCodebook watermarks recover short-text identifiers under selected edits: Authors report 99.9 percent exact recovery for 24-bit identifiers in one 200-token test setting, but certain instructed models and word-insertion edits substantially reduce recovery.
Results, 24-bit codebook and rewriting robustness tables
- supportsCodebook watermarks recover short-text identifiers under selected edits: A keyed identifier is mapped to a pseudorandom codeword, generation probabilities are biased to carry code bits and receivers use Hamming or Levenshtein-distance search to recover a likely identifier.
Methods, codeword generation and decoding strategy
- supportsCodebook watermarks recover short-text identifiers under selected edits: The paper reports that a 10 percent word-insertion edit reduces the Levenshtein codebook identifier match rate to roughly 60 percent, while base and instruction-tuned models differ materially in unedited recovery.
Section V-C Attack Robustness; word insertion experiment and model-family tables