{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:OVYMK562XE2E4WR2SW3GVIF2O2","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":"811fa6326ab2e562a174f72e5dd0e5825857b4e4dd4d63e546b42d1664c08e0b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-16T03:48:03Z","title_canon_sha256":"ec01ad32a6ebee1860a08defe268f753fe7244ef5bbcade7b3d4bb663cc3d1e9"},"schema_version":"1.0","source":{"id":"2507.11878","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.11878","created_at":"2026-07-07T02:18:25Z"},{"alias_kind":"arxiv_version","alias_value":"2507.11878v5","created_at":"2026-07-07T02:18:25Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.11878","created_at":"2026-07-07T02:18:25Z"},{"alias_kind":"pith_short_12","alias_value":"OVYMK562XE2E","created_at":"2026-07-07T02:18:25Z"},{"alias_kind":"pith_short_16","alias_value":"OVYMK562XE2E4WR2","created_at":"2026-07-07T02:18:25Z"},{"alias_kind":"pith_short_8","alias_value":"OVYMK562","created_at":"2026-07-07T02:18:25Z"}],"graph_snapshots":[{"event_id":"sha256:9c3ff2b5ece8c218dd44ed80d4cb4d101fd24bd9182c8f5efe2601e17cc01a98","target":"graph","created_at":"2026-07-07T02:18:25Z","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/2507.11878/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"LLMs are trained to refuse harmful instructions, but do they truly understand harmfulness beyond just refusing? Prior work has shown that LLMs' refusal behaviors can be mediated by a one-dimensional subspace, i.e., a refusal direction. In this work, we identify a new dimension to analyze safety mechanisms in LLMs, i.e., harmfulness, which is encoded internally as a separate concept from refusal. There exists a harmfulness direction that is distinct from the refusal direction. As causal evidence, steering along the harmfulness direction can lead LLMs to interpret harmless instructions as harmfu","authors_text":"David Bau, Jiachen Zhao, Jing Huang, Weiyan Shi, Zhengxuan Wu","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-16T03:48:03Z","title":"LLMs Encode Harmfulness and Refusal Separately"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.11878","kind":"arxiv","version":5},"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":"sha256:486be848dc5c443feb15f0f4bcb434e3eaef179c1e1a666355a68789af2ddf84","target":"record","created_at":"2026-07-07T02:18:25Z","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":"811fa6326ab2e562a174f72e5dd0e5825857b4e4dd4d63e546b42d1664c08e0b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-16T03:48:03Z","title_canon_sha256":"ec01ad32a6ebee1860a08defe268f753fe7244ef5bbcade7b3d4bb663cc3d1e9"},"schema_version":"1.0","source":{"id":"2507.11878","kind":"arxiv","version":5}},"canonical_sha256":"7570c577dab9344e5a3a95b66aa0ba76a14a19d32a69e1157f0d1a599f91ba19","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"7570c577dab9344e5a3a95b66aa0ba76a14a19d32a69e1157f0d1a599f91ba19","first_computed_at":"2026-07-07T02:18:25.900807Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-07T02:18:25.900807Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"x6CiWCNKII6hRZ/Ukrs/4AZRBbVw9xcOAGxUprlzUujhHFIz69y4cehigzcH9JVVsNyQxfMh3bKJR4YQqN5HBw==","signature_status":"signed_v1","signed_at":"2026-07-07T02:18:25.901498Z","signed_message":"canonical_sha256_bytes"},"source_id":"2507.11878","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:486be848dc5c443feb15f0f4bcb434e3eaef179c1e1a666355a68789af2ddf84","sha256:9c3ff2b5ece8c218dd44ed80d4cb4d101fd24bd9182c8f5efe2601e17cc01a98"],"state_sha256":"4e914527b047bc139ac908c9e74d3b7350e498fe18de785499b77fb9c22aae7d"}