{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:E34YGTAQEACWIYDSHZWJE2332Y","short_pith_number":"pith:E34YGTAQ","canonical_record":{"source":{"id":"2408.11491","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-08-21T10:01:34Z","cross_cats_sorted":[],"title_canon_sha256":"bda01f33776da1a0d4fbcc5b2039d883f43243dd6a670b9f5e1a9f1ad0f10e66","abstract_canon_sha256":"7181d56fe70c5821faa1acd859517727485e26e06b85bc622c4e8079bb73bb55"},"schema_version":"1.0"},"canonical_sha256":"26f9834c1020056460723e6c926b7bd62be63a5c5e98e3b6bb3df7e5b57d7d1c","source":{"kind":"arxiv","id":"2408.11491","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2408.11491","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"arxiv_version","alias_value":"2408.11491v2","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.11491","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"pith_short_12","alias_value":"E34YGTAQEACW","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"pith_short_16","alias_value":"E34YGTAQEACWIYDS","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"pith_short_8","alias_value":"E34YGTAQ","created_at":"2026-07-05T09:50:04Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:E34YGTAQEACWIYDSHZWJE2332Y","target":"record","payload":{"canonical_record":{"source":{"id":"2408.11491","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-08-21T10:01:34Z","cross_cats_sorted":[],"title_canon_sha256":"bda01f33776da1a0d4fbcc5b2039d883f43243dd6a670b9f5e1a9f1ad0f10e66","abstract_canon_sha256":"7181d56fe70c5821faa1acd859517727485e26e06b85bc622c4e8079bb73bb55"},"schema_version":"1.0"},"canonical_sha256":"26f9834c1020056460723e6c926b7bd62be63a5c5e98e3b6bb3df7e5b57d7d1c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:50:04.832137Z","signature_b64":"6opfKkqEw7Z8+L8b9bboMbDp8CqD5WvI2rLG1Y3e+mfzhuKxOBI6NepqItfBMOQdQ8TDVSDI0y3YoMGA0aQVDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"26f9834c1020056460723e6c926b7bd62be63a5c5e98e3b6bb3df7e5b57d7d1c","last_reissued_at":"2026-07-05T09:50:04.831611Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:50:04.831611Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2408.11491","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":"2026-07-05T09:50:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hg9ygeG341SYGyZ107f3fguJaRMhv2hVcvznTKTw15CZqjk5L//R/BzFeTD6cSnWmNqTXYtzgBt1GiqLGK9KBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T16:59:35.537908Z"},"content_sha256":"e9a2d2fca4876ffbb20b83038943e0986358cd485341ea5ec34f4be4ce5e06a4","schema_version":"1.0","event_id":"sha256:e9a2d2fca4876ffbb20b83038943e0986358cd485341ea5ec34f4be4ce5e06a4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:E34YGTAQEACWIYDSHZWJE2332Y","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"SCANS: Mitigating the Exaggerated Safety for LLMs via Safety-Conscious Activation Steering","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Hai Zhao, Yifei Yang, Zouying Cao","submitted_at":"2024-08-21T10:01:34Z","abstract_excerpt":"Safety alignment is indispensable for Large Language Models (LLMs) to defend threats from malicious instructions. However, recent researches reveal safety-aligned LLMs prone to reject benign queries due to the exaggerated safety issue, limiting their helpfulness. In this paper, we propose a Safety-Conscious Activation Steering (SCANS) method to mitigate the exaggerated safety concerns in aligned LLMs. First, SCANS extracts the refusal steering vectors within the activation space and utilizes vocabulary projection to anchor some specific safety-critical layers which influence model refusal beha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.11491","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/2408.11491/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":"2026-07-05T09:50:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hfb1pczGIliDI9Hg3KSstbz6CtsRIs3mZ94qkFbXfJXNdZSVQc1tS6oW3R1M+ATwQLe/5eQiHmbwq3otsOw6DQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T16:59:35.538516Z"},"content_sha256":"c2e7023600b606fb5ff01c82da36bb957402fc953aa13fdb8fc6571aaf6f9e11","schema_version":"1.0","event_id":"sha256:c2e7023600b606fb5ff01c82da36bb957402fc953aa13fdb8fc6571aaf6f9e11"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/E34YGTAQEACWIYDSHZWJE2332Y/bundle.json","state_url":"https://pith.science/pith/E34YGTAQEACWIYDSHZWJE2332Y/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/E34YGTAQEACWIYDSHZWJE2332Y/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":"2026-08-12T16:59:35Z","links":{"resolver":"https://pith.science/pith/E34YGTAQEACWIYDSHZWJE2332Y","bundle":"https://pith.science/pith/E34YGTAQEACWIYDSHZWJE2332Y/bundle.json","state":"https://pith.science/pith/E34YGTAQEACWIYDSHZWJE2332Y/state.json","well_known_bundle":"https://pith.science/.well-known/pith/E34YGTAQEACWIYDSHZWJE2332Y/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:E34YGTAQEACWIYDSHZWJE2332Y","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":"7181d56fe70c5821faa1acd859517727485e26e06b85bc622c4e8079bb73bb55","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-08-21T10:01:34Z","title_canon_sha256":"bda01f33776da1a0d4fbcc5b2039d883f43243dd6a670b9f5e1a9f1ad0f10e66"},"schema_version":"1.0","source":{"id":"2408.11491","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2408.11491","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"arxiv_version","alias_value":"2408.11491v2","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.11491","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"pith_short_12","alias_value":"E34YGTAQEACW","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"pith_short_16","alias_value":"E34YGTAQEACWIYDS","created_at":"2026-07-05T09:50:04Z"},{"alias_kind":"pith_short_8","alias_value":"E34YGTAQ","created_at":"2026-07-05T09:50:04Z"}],"graph_snapshots":[{"event_id":"sha256:c2e7023600b606fb5ff01c82da36bb957402fc953aa13fdb8fc6571aaf6f9e11","target":"graph","created_at":"2026-07-05T09:50:04Z","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/2408.11491/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Safety alignment is indispensable for Large Language Models (LLMs) to defend threats from malicious instructions. However, recent researches reveal safety-aligned LLMs prone to reject benign queries due to the exaggerated safety issue, limiting their helpfulness. In this paper, we propose a Safety-Conscious Activation Steering (SCANS) method to mitigate the exaggerated safety concerns in aligned LLMs. First, SCANS extracts the refusal steering vectors within the activation space and utilizes vocabulary projection to anchor some specific safety-critical layers which influence model refusal beha","authors_text":"Hai Zhao, Yifei Yang, Zouying Cao","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-08-21T10:01:34Z","title":"SCANS: Mitigating the Exaggerated Safety for LLMs via Safety-Conscious Activation Steering"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.11491","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:e9a2d2fca4876ffbb20b83038943e0986358cd485341ea5ec34f4be4ce5e06a4","target":"record","created_at":"2026-07-05T09:50:04Z","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":"7181d56fe70c5821faa1acd859517727485e26e06b85bc622c4e8079bb73bb55","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-08-21T10:01:34Z","title_canon_sha256":"bda01f33776da1a0d4fbcc5b2039d883f43243dd6a670b9f5e1a9f1ad0f10e66"},"schema_version":"1.0","source":{"id":"2408.11491","kind":"arxiv","version":2}},"canonical_sha256":"26f9834c1020056460723e6c926b7bd62be63a5c5e98e3b6bb3df7e5b57d7d1c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"26f9834c1020056460723e6c926b7bd62be63a5c5e98e3b6bb3df7e5b57d7d1c","first_computed_at":"2026-07-05T09:50:04.831611Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:50:04.831611Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6opfKkqEw7Z8+L8b9bboMbDp8CqD5WvI2rLG1Y3e+mfzhuKxOBI6NepqItfBMOQdQ8TDVSDI0y3YoMGA0aQVDA==","signature_status":"signed_v1","signed_at":"2026-07-05T09:50:04.832137Z","signed_message":"canonical_sha256_bytes"},"source_id":"2408.11491","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e9a2d2fca4876ffbb20b83038943e0986358cd485341ea5ec34f4be4ce5e06a4","sha256:c2e7023600b606fb5ff01c82da36bb957402fc953aa13fdb8fc6571aaf6f9e11"],"state_sha256":"56d4509d8e84b86bea819a6506524f6d49149027f4742225d4aecd784f5230fc"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"f5UMsPnUaCi886Tzqgqtztc+tpfRb/g7z7SQhVPt+Z1fiQ4fkfh7d+Ux+Is6shPQu1vDiiOmQ+WQNWX9/FE1AA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-12T16:59:35.542782Z","bundle_sha256":"74936dda34ae4e57c0b702d433423ca126c59185744ca141837ddc828a9bff31"}}