{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003446921,0.0009361619,0.0004939658,0.0005738047,0.0004908731,0.001065715,0.001219237,0.001263464,0.002544427],"category_scores_gemma":[0.01384345,0.0003019339,0.0007303798,0.0002624274,0.001444464,0.002006667,0.001763978,0.002260629,0.0006580561],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001547957,"about_ca_system_score_gemma":0.001677611,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004562905,"about_ca_topic_score_gemma":0.00481059,"domain_scores_codex":[0.9984202,0.0006249475,0.00007931077,0.0002218914,0.0004249609,0.0002287626],"domain_scores_gemma":[0.9928188,0.004281708,0.0005603568,0.001384209,0.0006243165,0.0003306257],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0005756299,0.0001415825,0.005330151,0.0001712497,0.00009178136,0.0001120535,0.0001106731,0.9566882,0.003542025,0.009993967,0.004934784,0.018308],"study_design_scores_gemma":[0.00002686269,0.0001679997,0.0003965333,0.00002355985,0.00001283793,0.00003302834,0.00003236284,0.9909822,0.003310172,0.004125188,0.0008780016,0.00001134596],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8478556,0.001620738,0.1226839,0.002821151,0.000368129,0.0002124511,0.001793626,0.008812902,0.0138314],"genre_scores_gemma":[0.9820247,0.0002548854,0.01437897,0.0002667538,0.00002310691,0.00007021453,0.0009631917,0.0003646154,0.001653729],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004562905,"threshold_uncertainty_score":0.01822931,"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."}}