{"id":"W4412888061","doi":"10.18653/v1/2025.findings-acl.784","title":"Enhancing LLM Agent Safety via Causal Influence Prompting","year":2025,"lang":"en","type":"article","venue":"","topic":"Software Engineering Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"Institute for Information and Communications Technology Promotion; Ministry of Science and ICT, South Korea; Korea Advanced Institute of Science and Technology","keywords":"Computer science; Human–computer interaction; Risk analysis (engineering); Business","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":[0.003685605,0.001480271,0.0006505796,0.001648364,0.0009022734,0.001442451,0.001531256,0.0009348014,0.005628103],"category_scores_gemma":[0.0243005,0.0007042626,0.00122317,0.0005799223,0.001439166,0.002801263,0.003836128,0.002080085,0.0008808732],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001228733,"about_ca_system_score_gemma":0.003516959,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004605404,"about_ca_topic_score_gemma":0.007360469,"domain_scores_codex":[0.9965628,0.001249642,0.0001487496,0.0005670458,0.001277661,0.0001940669],"domain_scores_gemma":[0.9877909,0.008159846,0.000937271,0.001481919,0.001292681,0.0003372574],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007545407,0.0004686706,0.009981304,0.0008474962,0.0001193914,0.00106299,0.003078362,0.3215854,0.03736548,0.1102877,0.007854068,0.5065946],"study_design_scores_gemma":[0.00007825371,0.0001329219,0.0004554304,0.00006947832,0.00008045397,0.0001746515,0.0002150505,0.8967539,0.03296549,0.04927089,0.01975323,0.00005031383],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01230613,0.0001058947,0.9775514,0.0002537513,0.00004309238,0.0001980279,0.0001154326,0.006832512,0.002593716],"genre_scores_gemma":[0.4164892,0.0001771472,0.5783402,0.0002298716,0.00004856398,0.0003323227,0.000405695,0.0009787168,0.00299819],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005628103,"threshold_uncertainty_score":0.01949161,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.009514871464068484,"score_gpt":0.2732855928382368,"score_spread":0.2637707213741683,"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."}}