{"id":"W4413219439","doi":"10.1007/s10664-025-10693-3","title":"Adversarial attack classification and robustness testing for large language models for code","year":2025,"lang":"en","type":"article","venue":"Empirical Software Engineering","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Computer science; Robustness (evolution); Adversarial system; Artificial intelligence; Programming language; Biology","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":[0.01019029,0.001596007,0.001372437,0.002321842,0.000833035,0.001825249,0.002147379,0.002283921,0.00305858],"category_scores_gemma":[0.06636024,0.0007081736,0.001791338,0.0009120548,0.002891566,0.003468318,0.004320963,0.004404186,0.0007008571],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001725834,"about_ca_system_score_gemma":0.001421004,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0027019,"about_ca_topic_score_gemma":0.002597441,"domain_scores_codex":[0.9926466,0.004241492,0.0003062542,0.00108961,0.001275224,0.000440767],"domain_scores_gemma":[0.9211485,0.06752677,0.002535941,0.006486217,0.001571986,0.000730628],"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":[0.0003641635,0.0002279033,0.004766596,0.0001381107,0.000190161,0.0001535202,0.0001747352,0.8483362,0.003438636,0.04318233,0.004530644,0.09449698],"study_design_scores_gemma":[0.000006371664,0.00001857192,0.0002160197,0.000006505561,0.00000578145,0.00001461936,0.000007422645,0.9843593,0.0005973116,0.01462859,0.0001338428,0.000005621049],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06454762,0.0004132369,0.9303793,0.001025274,0.00007119097,0.00008522037,0.0002318161,0.001718581,0.001527669],"genre_scores_gemma":[0.8882949,0.000220999,0.1063923,0.0003618463,0.0002159933,0.0001717199,0.001079498,0.0004454501,0.002817257],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01019029,"threshold_uncertainty_score":0.05389208,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04674644636670435,"score_gpt":0.3271197308610341,"score_spread":0.2803732844943297,"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."}}