{"id":"W4415965747","doi":"10.2139/ssrn.5705186","title":"Open Technical Problems in Open-Weight AI Model Risk Management","year":2025,"lang":"","type":"preprint","venue":"SSRN Electronic Journal","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Université de Montréal; McGill University; Mila - Quebec Artificial Intelligence Institute; University of Toronto; Institute on Governance","funders":"","keywords":"Key (lock); Risk management; Openness to experience; Risk assessment; Training (meteorology); Best practice","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":[0.009207043,0.001430969,0.002101679,0.001185306,0.001450665,0.005616416,0.003915006,0.004565706,0.01070924],"category_scores_gemma":[0.04795785,0.0008508677,0.001281686,0.001546045,0.005248223,0.0108073,0.006177265,0.009676292,0.001005516],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001840241,"about_ca_system_score_gemma":0.001531032,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001186843,"about_ca_topic_score_gemma":0.0006977158,"domain_scores_codex":[0.9949523,0.002238357,0.0002584729,0.0008266473,0.001411908,0.000312306],"domain_scores_gemma":[0.9633325,0.03031186,0.00141237,0.00221622,0.001918679,0.0008083375],"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":[0.00002552925,0.0000503388,0.0002346377,0.0001140333,0.00004270724,0.00005512451,0.00008217164,0.05131888,0.0002343957,0.9279374,0.00318333,0.01672149],"study_design_scores_gemma":[0.000004579988,0.00000836463,0.00004026772,0.00001476395,0.000004766673,0.00001346613,0.00001411052,0.1004201,0.00007827645,0.8986158,0.0007780301,0.000007445683],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01301549,0.00174956,0.9586912,0.009213116,0.0004331232,0.00003893713,0.0001319793,0.0001611489,0.01656537],"genre_scores_gemma":[0.7910485,0.00479125,0.1696524,0.002514988,0.004014628,0.0003816145,0.0005135067,0.0004389161,0.02664411],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01070924,"threshold_uncertainty_score":0.04869205,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01651247511547286,"score_gpt":0.313643105026537,"score_spread":0.2971306299110641,"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."}}