{"id":"W4409361367","doi":"10.1609/aaai.v39i26.34929","title":"Scaling Trends for Data Poisoning in LLMs","year":2025,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Institut de Valorisation des Données","keywords":"Scaling; Environmental health; Environmental science; Medicine; Mathematics","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":[0.007427972,0.0009474236,0.0007041634,0.001319561,0.0008524838,0.002020418,0.001696974,0.001300648,0.001684196],"category_scores_gemma":[0.05551858,0.0006410354,0.0009443914,0.0009436547,0.002204771,0.005207219,0.00308816,0.003685592,0.0007107256],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001182055,"about_ca_system_score_gemma":0.001214942,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004394704,"about_ca_topic_score_gemma":0.004044876,"domain_scores_codex":[0.9949307,0.001675837,0.0003824474,0.001138079,0.001476505,0.0003963347],"domain_scores_gemma":[0.9640507,0.01542114,0.002337367,0.01427453,0.002929399,0.0009868855],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001643267,0.0008711109,0.2042332,0.0007307006,0.0007111726,0.001124411,0.00238094,0.5497229,0.03339928,0.01518367,0.02833912,0.1616603],"study_design_scores_gemma":[0.0001388158,0.001117941,0.02175532,0.000120943,0.0001449726,0.0006497977,0.0009010086,0.913922,0.02848086,0.02025312,0.01242352,0.00009163899],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9336722,0.001385015,0.0430931,0.003705202,0.0002434855,0.0001517879,0.001247469,0.01141574,0.005085964],"genre_scores_gemma":[0.9814243,0.0001968944,0.01583913,0.0003919273,0.00002768118,0.00005844789,0.0009643858,0.0004169933,0.000680229],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007427972,"threshold_uncertainty_score":0.03928334,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.4474056700080657,"score_gpt":0.476595572378622,"score_spread":0.02918990237055635,"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."}}