{"id":"W4389736946","doi":"10.1561/3300000041","title":"Identifying and Mitigating the Security Risks of Generative AI","year":2023,"lang":"en","type":"article","venue":"Foundations and Trends® in Privacy and Security","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":46,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Generative grammar; Computer science; Business; Artificial intelligence","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":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.001297727,0.00007687807,0.0001367389,0.0001145834,0.001325751,0.0002902597,0.00009135256,0.00009461765,0.00003133648],"category_scores_gemma":[0.0005027708,0.00006426623,0.00002836563,0.0005239057,0.0006442831,0.0004309933,0.000125384,0.0002765266,7.837702e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001494707,"about_ca_system_score_gemma":0.00004760924,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00583177,"about_ca_topic_score_gemma":0.01329804,"domain_scores_codex":[0.9990394,0.000239351,0.000190264,0.0001692149,0.0001735695,0.0001882235],"domain_scores_gemma":[0.9992773,0.0003875313,0.00008190374,0.00009076048,0.00008986227,0.00007264531],"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":[0.00000583838,0.00004316951,0.03308604,0.00003598442,0.00003433741,0.000003139795,0.4356862,0.000001654026,0.00003950025,0.5105689,0.0004046996,0.02009056],"study_design_scores_gemma":[0.00036558,0.00003126285,0.1524967,0.00004865323,0.00002579639,0.00000101408,0.02606467,0.001077608,0.00004077449,0.8143025,0.005395006,0.0001504364],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9793484,0.0003269479,0.00001642024,0.01541311,0.00009493909,0.00008944924,0.00002456745,0.00002296753,0.004663185],"genre_scores_gemma":[0.997912,0.001666978,0.00006736282,0.0001612378,0.00007788412,0.000007440448,0.00001535815,0.000004074976,0.00008765321],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4096215,"threshold_uncertainty_score":0.9999744,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1558116364776305,"score_gpt":0.4603913893009867,"score_spread":0.3045797528233561,"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."}}