{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:LB6M4H2YOBUYZTQQJZFXAT43CI","short_pith_number":"pith:LB6M4H2Y","canonical_record":{"source":{"id":"2605.14074","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T19:50:35Z","cross_cats_sorted":[],"title_canon_sha256":"6474fa5dc1d69786de7cc7a9010ec3710929a3e99b89268e89646fbe05119328","abstract_canon_sha256":"3ecd350b56951bc7424beb6da781a938e660910527815ad3d5ee02497cd04d29"},"schema_version":"1.0"},"canonical_sha256":"587cce1f5870698cce104e4b704f9b1212246c162f98492efdd1460b0bfacb21","source":{"kind":"arxiv","id":"2605.14074","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14074","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14074v1","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14074","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"pith_short_12","alias_value":"LB6M4H2YOBUY","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"LB6M4H2YOBUYZTQQ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"LB6M4H2Y","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:LB6M4H2YOBUYZTQQJZFXAT43CI","target":"record","payload":{"canonical_record":{"source":{"id":"2605.14074","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T19:50:35Z","cross_cats_sorted":[],"title_canon_sha256":"6474fa5dc1d69786de7cc7a9010ec3710929a3e99b89268e89646fbe05119328","abstract_canon_sha256":"3ecd350b56951bc7424beb6da781a938e660910527815ad3d5ee02497cd04d29"},"schema_version":"1.0"},"canonical_sha256":"587cce1f5870698cce104e4b704f9b1212246c162f98492efdd1460b0bfacb21","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:12.382759Z","signature_b64":"aOrA0vZITgS43WfzhlpVtqwumZY/vI+uieSuxC2OGRAEjmUrdftwvw/pMw8Ba1Hk4G1ucmalhz38Nj59XGKvBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"587cce1f5870698cce104e4b704f9b1212246c162f98492efdd1460b0bfacb21","last_reissued_at":"2026-05-17T23:39:12.382333Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:12.382333Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.14074","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"59R87x7Ip4oAbh49Ko86zPtNowJGj0vfQN+L+MZad4bTRLBzLGGBsQtN3k9oV6ZR8gJeKuZLT2iJb4CD09rKCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T07:37:50.870393Z"},"content_sha256":"2035e727ffc74fc46485e1e7c96ce7b552510790c562420aa57653ca7a33fecc","schema_version":"1.0","event_id":"sha256:2035e727ffc74fc46485e1e7c96ce7b552510790c562420aa57653ca7a33fecc"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:LB6M4H2YOBUYZTQQJZFXAT43CI","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Fair and Calibrated Toxicity Detection with Robust Training and Abstention","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Toxicity detectors hide calibration unfairness across identity subgroups despite near-perfect overall scores.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Mokshit Surana","submitted_at":"2026-05-13T19:50:35Z","abstract_excerpt":"Fairness in toxicity classification involves three integrated axes: ranking, calibration, and abstention. Training-time interventions and post-hoc safety mechanisms cannot be evaluated independently because the former determines the efficacy of the latter. We compare Empirical Risk Minimization (ERM), instance-level reweighting, and Group DRO across these axes, combined with temperature scaling, confidence-based abstention, and per-identity threshold optimization. Evaluation uses subgroup AUC, BPSN/BNSP AUC, error gaps, and per-subgroup Expected Calibration Error (ECE) with bootstrap CIs ($n ="},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Calibration disparity is a hidden fairness violation. ERM has near-perfect aggregate calibration (0.013) but is significantly miscalibrated across all identity subgroups (+0.029 to +0.134). Training interventions reshape rather than eliminate disparity, and abstention itself is unfair.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the chosen subgroup definitions, metrics (subgroup AUC, BPSN/BNSP AUC, ECE), and bootstrap CIs fully capture real-world fairness harms and that post-hoc methods can be evaluated independently of training choices.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Training interventions reshape rather than eliminate calibration and abstention disparities in toxicity detection, requiring a multi-axis fairness framework.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Toxicity detectors hide calibration unfairness across identity subgroups despite near-perfect overall scores.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d5dbafe0cf106450938bf58f5b8e6ed37c21bb28e07fb256dd3d2486e4f4ed8e"},"source":{"id":"2605.14074","kind":"arxiv","version":1},"verdict":{"id":"489b6283-612b-4940-97b4-9db6692fba1e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T05:06:16.459603Z","strongest_claim":"Calibration disparity is a hidden fairness violation. ERM has near-perfect aggregate calibration (0.013) but is significantly miscalibrated across all identity subgroups (+0.029 to +0.134). Training interventions reshape rather than eliminate disparity, and abstention itself is unfair.","one_line_summary":"Training interventions reshape rather than eliminate calibration and abstention disparities in toxicity detection, requiring a multi-axis fairness framework.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the chosen subgroup definitions, metrics (subgroup AUC, BPSN/BNSP AUC, ECE), and bootstrap CIs fully capture real-world fairness harms and that post-hoc methods can be evaluated independently of training choices.","pith_extraction_headline":"Toxicity detectors hide calibration unfairness across identity subgroups despite near-perfect overall scores."},"references":{"count":11,"sample":[{"doi":"","year":2019,"title":"Borkan, D., Dixon, L., Sorensen, J., Thain, N., and Vasserman, L. (2019). Nuanced metrics for measuring unintended bias with real data for text classification.WWW Companion","work_id":"3caf62a1-4189-425f-a7df-ea89c0d1365a","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Dixon, L., Li, J., Sorensen, J., Thain, N., and Vasserman, L. (2018). Measuring and mitigating unintended bias in text classification.AAAI/ACM AIES","work_id":"55b0a4ff-f708-47f0-83c5-a5d676db6a39","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Geifman, Y ., and El-Yaniv, R. (2017). Selective classification for deep neural networks.NeurIPS 2017","work_id":"021ef3f4-3004-4ba0-bbdb-d05a03b7acbc","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Guo, C., Pleiss, G., Sun, Y ., and Weinberger, K. Q. (2017). On calibration of modern neural networks.ICML 2017","work_id":"c1da11f2-7c41-4a02-9ee0-58c0967b7a7f","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Y ., Arjovsky, M., Pezeshki, M., and Lopez-Paz, D","work_id":"2430ab6f-f9b0-43ac-a6ce-15eba2f4d7c3","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":11,"snapshot_sha256":"599498bc6f961f09f584cce49721295015273607d73c97fee8ca86f11f643391","internal_anchors":0},"formal_canon":{"evidence_count":1,"snapshot_sha256":"1f8a2ee6f87ae8a710e85312f7b8c0cfe251e854650c344287e394fd91e1ad8a"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"489b6283-612b-4940-97b4-9db6692fba1e"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ZVfNe8+WaapLjzqXpdEsV06cDcQjoJ7lSpUn+vT085c74UHgM+u9veBNoOtBfMSAzbYmIHGxviSqI14jwau/Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T07:37:50.870931Z"},"content_sha256":"408fabdb312de8c2dd3f1a361991e618df3861fa4154d7b0d52522caa4a7eae8","schema_version":"1.0","event_id":"sha256:408fabdb312de8c2dd3f1a361991e618df3861fa4154d7b0d52522caa4a7eae8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/LB6M4H2YOBUYZTQQJZFXAT43CI/bundle.json","state_url":"https://pith.science/pith/LB6M4H2YOBUYZTQQJZFXAT43CI/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/LB6M4H2YOBUYZTQQJZFXAT43CI/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-30T07:37:50Z","links":{"resolver":"https://pith.science/pith/LB6M4H2YOBUYZTQQJZFXAT43CI","bundle":"https://pith.science/pith/LB6M4H2YOBUYZTQQJZFXAT43CI/bundle.json","state":"https://pith.science/pith/LB6M4H2YOBUYZTQQJZFXAT43CI/state.json","well_known_bundle":"https://pith.science/.well-known/pith/LB6M4H2YOBUYZTQQJZFXAT43CI/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:LB6M4H2YOBUYZTQQJZFXAT43CI","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"3ecd350b56951bc7424beb6da781a938e660910527815ad3d5ee02497cd04d29","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T19:50:35Z","title_canon_sha256":"6474fa5dc1d69786de7cc7a9010ec3710929a3e99b89268e89646fbe05119328"},"schema_version":"1.0","source":{"id":"2605.14074","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14074","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14074v1","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14074","created_at":"2026-05-17T23:39:12Z"},{"alias_kind":"pith_short_12","alias_value":"LB6M4H2YOBUY","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"LB6M4H2YOBUYZTQQ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"LB6M4H2Y","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:408fabdb312de8c2dd3f1a361991e618df3861fa4154d7b0d52522caa4a7eae8","target":"graph","created_at":"2026-05-17T23:39:12Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Calibration disparity is a hidden fairness violation. ERM has near-perfect aggregate calibration (0.013) but is significantly miscalibrated across all identity subgroups (+0.029 to +0.134). Training interventions reshape rather than eliminate disparity, and abstention itself is unfair."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the chosen subgroup definitions, metrics (subgroup AUC, BPSN/BNSP AUC, ECE), and bootstrap CIs fully capture real-world fairness harms and that post-hoc methods can be evaluated independently of training choices."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Training interventions reshape rather than eliminate calibration and abstention disparities in toxicity detection, requiring a multi-axis fairness framework."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Toxicity detectors hide calibration unfairness across identity subgroups despite near-perfect overall scores."}],"snapshot_sha256":"d5dbafe0cf106450938bf58f5b8e6ed37c21bb28e07fb256dd3d2486e4f4ed8e"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"1f8a2ee6f87ae8a710e85312f7b8c0cfe251e854650c344287e394fd91e1ad8a"},"paper":{"abstract_excerpt":"Fairness in toxicity classification involves three integrated axes: ranking, calibration, and abstention. Training-time interventions and post-hoc safety mechanisms cannot be evaluated independently because the former determines the efficacy of the latter. We compare Empirical Risk Minimization (ERM), instance-level reweighting, and Group DRO across these axes, combined with temperature scaling, confidence-based abstention, and per-identity threshold optimization. Evaluation uses subgroup AUC, BPSN/BNSP AUC, error gaps, and per-subgroup Expected Calibration Error (ECE) with bootstrap CIs ($n =","authors_text":"Mokshit Surana","cross_cats":[],"headline":"Toxicity detectors hide calibration unfairness across identity subgroups despite near-perfect overall scores.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T19:50:35Z","title":"Fair and Calibrated Toxicity Detection with Robust Training and Abstention"},"references":{"count":11,"internal_anchors":0,"resolved_work":11,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Borkan, D., Dixon, L., Sorensen, J., Thain, N., and Vasserman, L. (2019). Nuanced metrics for measuring unintended bias with real data for text classification.WWW Companion","work_id":"3caf62a1-4189-425f-a7df-ea89c0d1365a","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Dixon, L., Li, J., Sorensen, J., Thain, N., and Vasserman, L. (2018). Measuring and mitigating unintended bias in text classification.AAAI/ACM AIES","work_id":"55b0a4ff-f708-47f0-83c5-a5d676db6a39","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Geifman, Y ., and El-Yaniv, R. (2017). Selective classification for deep neural networks.NeurIPS 2017","work_id":"021ef3f4-3004-4ba0-bbdb-d05a03b7acbc","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Guo, C., Pleiss, G., Sun, Y ., and Weinberger, K. Q. (2017). On calibration of modern neural networks.ICML 2017","work_id":"c1da11f2-7c41-4a02-9ee0-58c0967b7a7f","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Y ., Arjovsky, M., Pezeshki, M., and Lopez-Paz, D","work_id":"2430ab6f-f9b0-43ac-a6ce-15eba2f4d7c3","year":2022}],"snapshot_sha256":"599498bc6f961f09f584cce49721295015273607d73c97fee8ca86f11f643391"},"source":{"id":"2605.14074","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T05:06:16.459603Z","id":"489b6283-612b-4940-97b4-9db6692fba1e","model_set":{"reader":"grok-4.3"},"one_line_summary":"Training interventions reshape rather than eliminate calibration and abstention disparities in toxicity detection, requiring a multi-axis fairness framework.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Toxicity detectors hide calibration unfairness across identity subgroups despite near-perfect overall scores.","strongest_claim":"Calibration disparity is a hidden fairness violation. ERM has near-perfect aggregate calibration (0.013) but is significantly miscalibrated across all identity subgroups (+0.029 to +0.134). Training interventions reshape rather than eliminate disparity, and abstention itself is unfair.","weakest_assumption":"That the chosen subgroup definitions, metrics (subgroup AUC, BPSN/BNSP AUC, ECE), and bootstrap CIs fully capture real-world fairness harms and that post-hoc methods can be evaluated independently of training choices."}},"verdict_id":"489b6283-612b-4940-97b4-9db6692fba1e"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:2035e727ffc74fc46485e1e7c96ce7b552510790c562420aa57653ca7a33fecc","target":"record","created_at":"2026-05-17T23:39:12Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"3ecd350b56951bc7424beb6da781a938e660910527815ad3d5ee02497cd04d29","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T19:50:35Z","title_canon_sha256":"6474fa5dc1d69786de7cc7a9010ec3710929a3e99b89268e89646fbe05119328"},"schema_version":"1.0","source":{"id":"2605.14074","kind":"arxiv","version":1}},"canonical_sha256":"587cce1f5870698cce104e4b704f9b1212246c162f98492efdd1460b0bfacb21","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"587cce1f5870698cce104e4b704f9b1212246c162f98492efdd1460b0bfacb21","first_computed_at":"2026-05-17T23:39:12.382333Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:12.382333Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"aOrA0vZITgS43WfzhlpVtqwumZY/vI+uieSuxC2OGRAEjmUrdftwvw/pMw8Ba1Hk4G1ucmalhz38Nj59XGKvBQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:12.382759Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14074","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:2035e727ffc74fc46485e1e7c96ce7b552510790c562420aa57653ca7a33fecc","sha256:408fabdb312de8c2dd3f1a361991e618df3861fa4154d7b0d52522caa4a7eae8"],"state_sha256":"cd5c9d36425e8ca356c2c2a9b30b705037594ad0a4a4e8b377f30af9c565051b"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"MDOK6S7pUiCedQGnTbxncBcYiQOZa0+/dWo0MjlVnCEoB2+wYYNx03r0Yqddj1EcUgtR3DXNK159U80tR2E9Ag==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-30T07:37:50.873543Z","bundle_sha256":"452e560a039e43519c0eff63acb29e3b73a7cb2f5d22267751037e03ffbc73cd"}}