{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003415935,0.0007185633,0.0006417558,0.001021008,0.001249568,0.004445342,0.001509733,0.002415664,0.005537525],"category_scores_gemma":[0.02336537,0.0006174555,0.001510849,0.0007471741,0.004379543,0.01113711,0.004516277,0.0037153,0.001567005],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002626303,"about_ca_system_score_gemma":0.002028486,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007070064,"about_ca_topic_score_gemma":0.0007093837,"domain_scores_codex":[0.9947885,0.001475055,0.0003371088,0.0007127778,0.002122331,0.0005641796],"domain_scores_gemma":[0.97939,0.01007746,0.00236616,0.006579208,0.001191748,0.0003954592],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00009913353,0.00005550793,0.0002963735,0.0001003683,0.00002012783,0.0002841623,0.0003834009,0.02040892,0.006885094,0.9500797,0.001556157,0.01983107],"study_design_scores_gemma":[0.00003602435,0.00006096856,0.00008866499,0.00006648825,0.00002358575,0.0002604253,0.00005910993,0.2067746,0.01558325,0.7699472,0.007044728,0.00005511334],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02569828,0.0002927476,0.9600486,0.001993128,0.00009634438,0.00006908891,0.0001987164,0.001577833,0.01002522],"genre_scores_gemma":[0.8567418,0.0004386331,0.133518,0.0005942034,0.0003312886,0.0001668831,0.0003395434,0.000386277,0.007483282],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005537525,"threshold_uncertainty_score":0.01905525,"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."}}