{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:3E63MUJ4HYR2BFVQF4CMFLOTKQ","short_pith_number":"pith:3E63MUJ4","canonical_record":{"source":{"id":"2405.18822","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-29T07:03:31Z","cross_cats_sorted":[],"title_canon_sha256":"f803457a8c6a88f48839c7303ecc60b706269d40b123153616461b974c1f6a2d","abstract_canon_sha256":"6f217941a76817f6224d63cd762a6c542f3461f57df67fe06bb302fa2c9f385d"},"schema_version":"1.0"},"canonical_sha256":"d93db6513c3e23a096b02f04c2add3541d9003e5019e55943b3bb591c1187106","source":{"kind":"arxiv","id":"2405.18822","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.18822","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"arxiv_version","alias_value":"2405.18822v2","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18822","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"pith_short_12","alias_value":"3E63MUJ4HYR2","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"pith_short_16","alias_value":"3E63MUJ4HYR2BFVQ","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"pith_short_8","alias_value":"3E63MUJ4","created_at":"2026-07-05T09:32:44Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:3E63MUJ4HYR2BFVQF4CMFLOTKQ","target":"record","payload":{"canonical_record":{"source":{"id":"2405.18822","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-29T07:03:31Z","cross_cats_sorted":[],"title_canon_sha256":"f803457a8c6a88f48839c7303ecc60b706269d40b123153616461b974c1f6a2d","abstract_canon_sha256":"6f217941a76817f6224d63cd762a6c542f3461f57df67fe06bb302fa2c9f385d"},"schema_version":"1.0"},"canonical_sha256":"d93db6513c3e23a096b02f04c2add3541d9003e5019e55943b3bb591c1187106","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:32:44.414742Z","signature_b64":"afBhW6ptC4ekbQ7ys0+wbSXByl/Jc6qUTQbTI15AmbZ3jPqw1XfHFu73BLGKHjMDfjs0xkz5xkJJchHGvSQ7Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d93db6513c3e23a096b02f04c2add3541d9003e5019e55943b3bb591c1187106","last_reissued_at":"2026-07-05T09:32:44.414240Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:32:44.414240Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.18822","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:32:44Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"dK4+hHSeNlypkhgbP3VQT6IIH3xJYaswIhJcLyuEUeJphQJ0VxYw25U/rw56dAaSTp4iFA23r0zxN5yphOOJCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T18:20:36.356063Z"},"content_sha256":"a28fd43d172d34970a28113dbb9f31785c13d57156e52b2e634718f67689bc9c","schema_version":"1.0","event_id":"sha256:a28fd43d172d34970a28113dbb9f31785c13d57156e52b2e634718f67689bc9c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:3E63MUJ4HYR2BFVQF4CMFLOTKQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Toxicity Detection for Free","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"David Wagner, Geng Zhao, Jiantao Jiao, Julien Piet, Zhanhao Hu","submitted_at":"2024-05-29T07:03:31Z","abstract_excerpt":"Current LLMs are generally aligned to follow safety requirements and tend to refuse toxic prompts. However, LLMs can fail to refuse toxic prompts or be overcautious and refuse benign examples. In addition, state-of-the-art toxicity detectors have low TPRs at low FPR, incurring high costs in real-world applications where toxic examples are rare. In this paper, we introduce Moderation Using LLM Introspection (MULI), which detects toxic prompts using the information extracted directly from LLMs themselves. We found we can distinguish between benign and toxic prompts from the distribution of the f"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18822","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2405.18822/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:32:44Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"woa38dC+bm5C7r6C2RhbDOy375P3OMEGKoWQU4RBuINzIeA9s35KKQCps1aSXtTHHzKn/ESSImA3oVC9zjdFDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T18:20:36.356829Z"},"content_sha256":"114dfe9bf3782e8ed8937ef5381ccb085e9b6f87f46400865137465b20363309","schema_version":"1.0","event_id":"sha256:114dfe9bf3782e8ed8937ef5381ccb085e9b6f87f46400865137465b20363309"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3E63MUJ4HYR2BFVQF4CMFLOTKQ/bundle.json","state_url":"https://pith.science/pith/3E63MUJ4HYR2BFVQF4CMFLOTKQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3E63MUJ4HYR2BFVQF4CMFLOTKQ/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":"2026-08-08T18:20:36Z","links":{"resolver":"https://pith.science/pith/3E63MUJ4HYR2BFVQF4CMFLOTKQ","bundle":"https://pith.science/pith/3E63MUJ4HYR2BFVQF4CMFLOTKQ/bundle.json","state":"https://pith.science/pith/3E63MUJ4HYR2BFVQF4CMFLOTKQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3E63MUJ4HYR2BFVQF4CMFLOTKQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:3E63MUJ4HYR2BFVQF4CMFLOTKQ","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":"6f217941a76817f6224d63cd762a6c542f3461f57df67fe06bb302fa2c9f385d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-29T07:03:31Z","title_canon_sha256":"f803457a8c6a88f48839c7303ecc60b706269d40b123153616461b974c1f6a2d"},"schema_version":"1.0","source":{"id":"2405.18822","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.18822","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"arxiv_version","alias_value":"2405.18822v2","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18822","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"pith_short_12","alias_value":"3E63MUJ4HYR2","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"pith_short_16","alias_value":"3E63MUJ4HYR2BFVQ","created_at":"2026-07-05T09:32:44Z"},{"alias_kind":"pith_short_8","alias_value":"3E63MUJ4","created_at":"2026-07-05T09:32:44Z"}],"graph_snapshots":[{"event_id":"sha256:114dfe9bf3782e8ed8937ef5381ccb085e9b6f87f46400865137465b20363309","target":"graph","created_at":"2026-07-05T09:32:44Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2405.18822/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Current LLMs are generally aligned to follow safety requirements and tend to refuse toxic prompts. However, LLMs can fail to refuse toxic prompts or be overcautious and refuse benign examples. In addition, state-of-the-art toxicity detectors have low TPRs at low FPR, incurring high costs in real-world applications where toxic examples are rare. In this paper, we introduce Moderation Using LLM Introspection (MULI), which detects toxic prompts using the information extracted directly from LLMs themselves. We found we can distinguish between benign and toxic prompts from the distribution of the f","authors_text":"David Wagner, Geng Zhao, Jiantao Jiao, Julien Piet, Zhanhao Hu","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-29T07:03:31Z","title":"Toxicity Detection for Free"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18822","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:a28fd43d172d34970a28113dbb9f31785c13d57156e52b2e634718f67689bc9c","target":"record","created_at":"2026-07-05T09:32:44Z","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":"6f217941a76817f6224d63cd762a6c542f3461f57df67fe06bb302fa2c9f385d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-29T07:03:31Z","title_canon_sha256":"f803457a8c6a88f48839c7303ecc60b706269d40b123153616461b974c1f6a2d"},"schema_version":"1.0","source":{"id":"2405.18822","kind":"arxiv","version":2}},"canonical_sha256":"d93db6513c3e23a096b02f04c2add3541d9003e5019e55943b3bb591c1187106","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d93db6513c3e23a096b02f04c2add3541d9003e5019e55943b3bb591c1187106","first_computed_at":"2026-07-05T09:32:44.414240Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:32:44.414240Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"afBhW6ptC4ekbQ7ys0+wbSXByl/Jc6qUTQbTI15AmbZ3jPqw1XfHFu73BLGKHjMDfjs0xkz5xkJJchHGvSQ7Dg==","signature_status":"signed_v1","signed_at":"2026-07-05T09:32:44.414742Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.18822","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a28fd43d172d34970a28113dbb9f31785c13d57156e52b2e634718f67689bc9c","sha256:114dfe9bf3782e8ed8937ef5381ccb085e9b6f87f46400865137465b20363309"],"state_sha256":"4d1aa891dbdeaf504b8af800e31f737fc6912ff3aad0cbd65e7a9058eadab62e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"IdNus8gXiYU2tCj1dZwm58+4mY+Yzx9+CnUFt56+RSvfiDur2Km2uz37DKycoDbCS8tD4ijNYF3efE3XaVZMAg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T18:20:36.362168Z","bundle_sha256":"ba43c9ddbdef1dd45557018252b083c0af53674e59d362973c1d29819ded4340"}}