{"id":"W7125580424","doi":"10.1109/cascon66301.2025.00121","title":"Evaluating Toxicity Understanding of LLM Agents","year":2025,"lang":"","type":"article","venue":"","topic":"Hate Speech and Cyberbullying Detection","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Toxicity; Perception; Trustworthiness; Order (exchange); Action (physics); Risk assessment","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":[0.0133151,0.001072314,0.0006088388,0.002339478,0.001273681,0.004362482,0.001084165,0.002318359,0.003117309],"category_scores_gemma":[0.08940034,0.0003768178,0.0006648328,0.0005760761,0.002480272,0.007274328,0.005093307,0.001709542,0.0009153582],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003031352,"about_ca_system_score_gemma":0.001526519,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00347947,"about_ca_topic_score_gemma":0.002805075,"domain_scores_codex":[0.9850186,0.007684497,0.00105065,0.001218924,0.00458448,0.0004428911],"domain_scores_gemma":[0.9456544,0.02993513,0.01168231,0.003867755,0.007656951,0.001203479],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.002225132,0.001152818,0.3209711,0.002147914,0.000515811,0.001786208,0.1134501,0.04587957,0.04876775,0.04985751,0.007474965,0.4057712],"study_design_scores_gemma":[0.0001437104,0.005812382,0.3020931,0.00166661,0.0008786684,0.002189814,0.07713888,0.2675637,0.1178803,0.1234488,0.1004779,0.0007061714],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8234109,0.001049348,0.1287995,0.002790097,0.00007131546,0.0007106328,0.0005044391,0.00137457,0.04128915],"genre_scores_gemma":[0.9615948,0.0003391178,0.03451382,0.0003797809,0.00003099189,0.000143833,0.0005560569,0.0001023349,0.002339315],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.0133151,"threshold_uncertainty_score":0.07041782,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1342825005152352,"score_gpt":0.3670419346009326,"score_spread":0.2327594340856974,"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."}}