{"id":"W4406309654","doi":"10.62056/ahmpdkp10","title":"Publicly-Detectable Watermarking for Language Models","year":2025,"lang":"en","type":"article","venue":"IACR Communications in Cryptology","topic":"Advanced Steganography and Watermarking Techniques","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Army Research Office; Air Force Office of Scientific Research; National Science Foundation","keywords":"Digital watermarking; Computer science; Computer security; Artificial intelligence","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004169911,0.0001134908,0.0001906997,0.0004754039,0.0002559025,0.00008167041,0.00297096,0.0001130293,0.000001301746],"category_scores_gemma":[0.00005732831,0.0001140831,0.00006970693,0.0006655577,0.0001332028,0.0005432576,0.00103063,0.0002188297,0.000001357315],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004948429,"about_ca_system_score_gemma":0.00004611008,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005454247,"about_ca_topic_score_gemma":0.0001807284,"domain_scores_codex":[0.99891,0.0001565914,0.0002950948,0.0002732228,0.00005222948,0.0003128751],"domain_scores_gemma":[0.9971159,0.0004358315,0.00006963676,0.002277439,0.00007789321,0.00002336304],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000006653602,0.00005917896,0.0007154365,0.0000151696,0.00001168902,0.00000104314,0.0007406894,0.00008788688,0.0005637939,0.9600588,0.0002871604,0.03745247],"study_design_scores_gemma":[0.0003773082,0.00002546085,0.000171038,0.00005128069,0.000004660838,0.000006354173,0.00009446205,0.1034792,0.004368468,0.8641597,0.02710267,0.0001593333],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.002268614,0.001042559,0.9779776,0.004461213,0.0001314824,0.0003261316,0.000003684318,0.0003685527,0.0134202],"genre_scores_gemma":[0.5322467,0.0001276475,0.466949,0.0003281939,0.000004316785,0.0002983966,0.000008429976,0.000004588704,0.00003264458],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5299781,"threshold_uncertainty_score":0.5520833,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03659632076982125,"score_gpt":0.3409854020883518,"score_spread":0.3043890813185306,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}