{"id":"W6979244033","doi":"","title":"Security Assessment of DeepSeek and GPT Series Models against Jailbreak Attacks","year":2025,"lang":"en","type":"article","venue":"ArXiv.org","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Nvidia","keywords":"Robustness (evolution); Software deployment; Adversarial system; Modular design; Threat model; Vulnerability (computing); Vulnerability assessment","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003703466,0.000172012,0.0002723704,0.0001166443,0.0001741404,0.00006261391,0.0006159827,0.00008592921,0.000005412183],"category_scores_gemma":[0.0000794038,0.0001691398,0.00005679975,0.0003900469,0.0001289424,0.0008416045,0.0008074541,0.0003008668,0.00000305398],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005167303,"about_ca_system_score_gemma":0.0001378774,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005432004,"about_ca_topic_score_gemma":0.00002248464,"domain_scores_codex":[0.9986816,0.0001257418,0.0002934573,0.0004280377,0.0002161473,0.0002549876],"domain_scores_gemma":[0.9989881,0.0001446323,0.0001360517,0.0005593529,0.0001073585,0.00006450419],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001685016,0.0001166483,0.7130324,0.0001703055,0.0001071392,0.00003302231,0.001666958,0.07532062,0.001032221,0.2005333,0.0002919192,0.00767866],"study_design_scores_gemma":[0.0007320061,0.00008330135,0.2295062,0.0001297392,0.00002819324,0.000007365503,0.0002355353,0.7487261,0.001102058,0.01804638,0.001070247,0.0003327954],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5522074,0.000100499,0.4391347,0.0008313013,0.0002457829,0.0001045465,0.000001576263,0.0000936673,0.007280615],"genre_scores_gemma":[0.9773764,0.00005497724,0.0219318,0.0003056387,0.00003223644,0.00001051272,0.000002902544,0.000009042718,0.0002764799],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6734055,"threshold_uncertainty_score":0.6897322,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0182755333581578,"score_gpt":0.2967100467162475,"score_spread":0.2784345133580897,"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."}}