{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:G3FPC7243ZVK2AACD3I7VUMYAA","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":"c27f1f6bcdadef19657c2e4e092faca536ce99691368ad4abaaab98a8f55ea81","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-07-21T22:50:11Z","title_canon_sha256":"c4c74a7bab17b29fcb8c3600147e34b7e570668d7fd1f0b8a6271823cf0be834"},"schema_version":"1.0","source":{"id":"2407.15286","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.15286","created_at":"2026-07-05T09:17:14Z"},{"alias_kind":"arxiv_version","alias_value":"2407.15286v3","created_at":"2026-07-05T09:17:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.15286","created_at":"2026-07-05T09:17:14Z"},{"alias_kind":"pith_short_12","alias_value":"G3FPC7243ZVK","created_at":"2026-07-05T09:17:14Z"},{"alias_kind":"pith_short_16","alias_value":"G3FPC7243ZVK2AAC","created_at":"2026-07-05T09:17:14Z"},{"alias_kind":"pith_short_8","alias_value":"G3FPC724","created_at":"2026-07-05T09:17:14Z"}],"graph_snapshots":[{"event_id":"sha256:a2ff6ff79c99a744c38f4e36f521cc32809ad21a79cdc9b35e37754e222aa042","target":"graph","created_at":"2026-07-05T09:17:14Z","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/2407.15286/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) are capable of producing content that perpetuates stereotypes, discrimination, and toxicity. The recently proposed moral self-correction is a computationally efficient method for reducing harmful content in the responses of LLMs. However, the process of how injecting self-correction instructions can modify the behavior of LLMs remains under-explored. In this paper, we explore the effectiveness of moral self-correction by answering three research questions: (1) In what scenarios does moral self-correction work? (2) What are the internal mechanisms of LLMs, e.g., hid","authors_text":"Guangliang Liu, Haitao Mao, Jiliang Tang, Kristen Marie Johnson","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-07-21T22:50:11Z","title":"Intrinsic Self-correction for Enhanced Morality: An Analysis of Internal Mechanisms and the Superficial Hypothesis"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.15286","kind":"arxiv","version":3},"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":"sha256:922374be1595648216a7555d4a88f60f81ee30fce494fc48b284a3555a0a6ceb","target":"record","created_at":"2026-07-05T09:17:14Z","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":"c27f1f6bcdadef19657c2e4e092faca536ce99691368ad4abaaab98a8f55ea81","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-07-21T22:50:11Z","title_canon_sha256":"c4c74a7bab17b29fcb8c3600147e34b7e570668d7fd1f0b8a6271823cf0be834"},"schema_version":"1.0","source":{"id":"2407.15286","kind":"arxiv","version":3}},"canonical_sha256":"36caf17f5cde6aad00021ed1fad1980009066f95d496783aa9944a36005e1367","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"36caf17f5cde6aad00021ed1fad1980009066f95d496783aa9944a36005e1367","first_computed_at":"2026-07-05T09:17:14.703020Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:17:14.703020Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"EFq2QVDTeKhU5brrdao7/7VF4UAarghgZVHDWgkKY/rNJeHDctwGPtXVKxv3EKHSwD/iVLznx6MDhbk5Gw1wAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T09:17:14.703441Z","signed_message":"canonical_sha256_bytes"},"source_id":"2407.15286","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:922374be1595648216a7555d4a88f60f81ee30fce494fc48b284a3555a0a6ceb","sha256:a2ff6ff79c99a744c38f4e36f521cc32809ad21a79cdc9b35e37754e222aa042"],"state_sha256":"5f83073f6e98c532a39b078159864bc912f88a875510dfc51e4fde7b5481df66"}