{"id":"W4417299029","doi":"10.48550/arxiv.2503.15772","title":"Detecting LLM-Generated Peer Reviews","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Spam and Phishing Detection","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Institute for Catastrophic Loss Reduction; National Science Foundation","keywords":"Digital watermarking; Watermark; Covert; Implementation; Embedding; Statistical power; Statistical hypothesis testing; Word error rate; Type I and type II errors","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.03820739,0.001131145,0.001754787,0.003355284,0.001256283,0.004231147,0.002984884,0.002896921,0.002233884],"category_scores_gemma":[0.3121547,0.0009633064,0.0007779563,0.001827805,0.001574086,0.004473325,0.004344871,0.002181258,0.002648536],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001326535,"about_ca_system_score_gemma":0.003177168,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005442579,"about_ca_topic_score_gemma":0.0006740565,"domain_scores_codex":[0.9111682,0.03496979,0.005661009,0.009732218,0.03702601,0.001442806],"domain_scores_gemma":[0.5411866,0.2277744,0.09352652,0.0640258,0.06863018,0.00485655],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002230225,0.0004181694,0.09562017,0.003331836,0.000692763,0.004386877,0.006203888,0.02389459,0.06930853,0.06454919,0.03415195,0.6952119],"study_design_scores_gemma":[0.000490921,0.002398996,0.03032722,0.0009586949,0.0004705194,0.009337161,0.00177784,0.5607348,0.2054469,0.08710515,0.1004413,0.0005104771],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2320386,0.003227828,0.7330438,0.003741173,0.001294018,0.002293133,0.001188503,0.01125476,0.01191813],"genre_scores_gemma":[0.7370733,0.000627752,0.2531478,0.001205914,0.0005280032,0.0007040365,0.000623051,0.0004905494,0.005599726],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9617926,"threshold_uncertainty_score":0.2020624,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09076130526028045,"score_gpt":0.3050084791668169,"score_spread":0.2142471739065364,"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."}}