{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:DQG2QJARJMR4O6WDQ5BNPE6GO6","short_pith_number":"pith:DQG2QJAR","canonical_record":{"source":{"id":"2402.13494","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-21T03:09:21Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"aeb8b5b07a5d06f7518340c4a3a74c8b29b7279769945240b7dfce8d8b4b983a","abstract_canon_sha256":"cd4941353ae3525e2712aaf8d7daa65551754ae164ee44d669852b153d6a2f01"},"schema_version":"1.0"},"canonical_sha256":"1c0da824114b23c77ac38742d793c6778647dc7d8d65e38b5ad91909434b0824","source":{"kind":"arxiv","id":"2402.13494","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.13494","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"arxiv_version","alias_value":"2402.13494v2","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.13494","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"pith_short_12","alias_value":"DQG2QJARJMR4","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"pith_short_16","alias_value":"DQG2QJARJMR4O6WD","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"pith_short_8","alias_value":"DQG2QJAR","created_at":"2026-07-05T08:25:00Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:DQG2QJARJMR4O6WDQ5BNPE6GO6","target":"record","payload":{"canonical_record":{"source":{"id":"2402.13494","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-21T03:09:21Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"aeb8b5b07a5d06f7518340c4a3a74c8b29b7279769945240b7dfce8d8b4b983a","abstract_canon_sha256":"cd4941353ae3525e2712aaf8d7daa65551754ae164ee44d669852b153d6a2f01"},"schema_version":"1.0"},"canonical_sha256":"1c0da824114b23c77ac38742d793c6778647dc7d8d65e38b5ad91909434b0824","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:25:00.823131Z","signature_b64":"/43wk4nlwaWaTWW6CWfPEntQfhQKIlYOqhdsx8ndsThmpTE5rbBfdDRqn2jWUwLfjt/1LsRVKAcEy/YQFBNmCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1c0da824114b23c77ac38742d793c6778647dc7d8d65e38b5ad91909434b0824","last_reissued_at":"2026-07-05T08:25:00.822692Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:25:00.822692Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2402.13494","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":"2026-07-05T08:25:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tt9qKfz6bgNFYspneUKExRHAbxd/6BzRaFuW3NDIm1KnXo3t/1px9tYdJ0hW+mrxJ9KCan5JG6tfs/Asl1PYCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-13T08:41:52.207507Z"},"content_sha256":"878976dd53a92cf3bc0fa4f0e43d5a37e641a2d99d84536723c4c6e0b27fabfb","schema_version":"1.0","event_id":"sha256:878976dd53a92cf3bc0fa4f0e43d5a37e641a2d99d84536723c4c6e0b27fabfb"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:DQG2QJARJMR4O6WDQ5BNPE6GO6","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"GradSafe: Detecting Jailbreak Prompts for LLMs via Safety-Critical Gradient Analysis","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CL","authors_text":"Minghong Fang, Neil Gong, Renjie Pi, Yueqi Xie","submitted_at":"2024-02-21T03:09:21Z","abstract_excerpt":"Large Language Models (LLMs) face threats from jailbreak prompts. Existing methods for detecting jailbreak prompts are primarily online moderation APIs or finetuned LLMs. These strategies, however, often require extensive and resource-intensive data collection and training processes. In this study, we propose GradSafe, which effectively detects jailbreak prompts by scrutinizing the gradients of safety-critical parameters in LLMs. Our method is grounded in a pivotal observation: the gradients of an LLM's loss for jailbreak prompts paired with compliance response exhibit similar patterns on cert"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.13494","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/2402.13494/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":"2026-07-05T08:25:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hPtTAuwB4InjeIq+gQQTWTYGjntZudl//Dm4ufwbFSdoLwCoyYk74pNMiMg6/sfy/FzSyy6CqXwP/1crMJ9/Ag==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-13T08:41:52.207886Z"},"content_sha256":"b36b9ce4bea49f6b6d561eb9575a6c94a67ddfc3b799221519493643fdfa145e","schema_version":"1.0","event_id":"sha256:b36b9ce4bea49f6b6d561eb9575a6c94a67ddfc3b799221519493643fdfa145e"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DQG2QJARJMR4O6WDQ5BNPE6GO6/bundle.json","state_url":"https://pith.science/pith/DQG2QJARJMR4O6WDQ5BNPE6GO6/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DQG2QJARJMR4O6WDQ5BNPE6GO6/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":"2026-08-13T08:41:52Z","links":{"resolver":"https://pith.science/pith/DQG2QJARJMR4O6WDQ5BNPE6GO6","bundle":"https://pith.science/pith/DQG2QJARJMR4O6WDQ5BNPE6GO6/bundle.json","state":"https://pith.science/pith/DQG2QJARJMR4O6WDQ5BNPE6GO6/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DQG2QJARJMR4O6WDQ5BNPE6GO6/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:DQG2QJARJMR4O6WDQ5BNPE6GO6","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":"cd4941353ae3525e2712aaf8d7daa65551754ae164ee44d669852b153d6a2f01","cross_cats_sorted":["cs.CR"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-21T03:09:21Z","title_canon_sha256":"aeb8b5b07a5d06f7518340c4a3a74c8b29b7279769945240b7dfce8d8b4b983a"},"schema_version":"1.0","source":{"id":"2402.13494","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.13494","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"arxiv_version","alias_value":"2402.13494v2","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.13494","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"pith_short_12","alias_value":"DQG2QJARJMR4","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"pith_short_16","alias_value":"DQG2QJARJMR4O6WD","created_at":"2026-07-05T08:25:00Z"},{"alias_kind":"pith_short_8","alias_value":"DQG2QJAR","created_at":"2026-07-05T08:25:00Z"}],"graph_snapshots":[{"event_id":"sha256:b36b9ce4bea49f6b6d561eb9575a6c94a67ddfc3b799221519493643fdfa145e","target":"graph","created_at":"2026-07-05T08:25:00Z","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/2402.13494/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) face threats from jailbreak prompts. Existing methods for detecting jailbreak prompts are primarily online moderation APIs or finetuned LLMs. These strategies, however, often require extensive and resource-intensive data collection and training processes. In this study, we propose GradSafe, which effectively detects jailbreak prompts by scrutinizing the gradients of safety-critical parameters in LLMs. Our method is grounded in a pivotal observation: the gradients of an LLM's loss for jailbreak prompts paired with compliance response exhibit similar patterns on cert","authors_text":"Minghong Fang, Neil Gong, Renjie Pi, Yueqi Xie","cross_cats":["cs.CR"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-21T03:09:21Z","title":"GradSafe: Detecting Jailbreak Prompts for LLMs via Safety-Critical Gradient Analysis"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.13494","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":"sha256:878976dd53a92cf3bc0fa4f0e43d5a37e641a2d99d84536723c4c6e0b27fabfb","target":"record","created_at":"2026-07-05T08:25:00Z","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":"cd4941353ae3525e2712aaf8d7daa65551754ae164ee44d669852b153d6a2f01","cross_cats_sorted":["cs.CR"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-21T03:09:21Z","title_canon_sha256":"aeb8b5b07a5d06f7518340c4a3a74c8b29b7279769945240b7dfce8d8b4b983a"},"schema_version":"1.0","source":{"id":"2402.13494","kind":"arxiv","version":2}},"canonical_sha256":"1c0da824114b23c77ac38742d793c6778647dc7d8d65e38b5ad91909434b0824","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1c0da824114b23c77ac38742d793c6778647dc7d8d65e38b5ad91909434b0824","first_computed_at":"2026-07-05T08:25:00.822692Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:25:00.822692Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"/43wk4nlwaWaTWW6CWfPEntQfhQKIlYOqhdsx8ndsThmpTE5rbBfdDRqn2jWUwLfjt/1LsRVKAcEy/YQFBNmCg==","signature_status":"signed_v1","signed_at":"2026-07-05T08:25:00.823131Z","signed_message":"canonical_sha256_bytes"},"source_id":"2402.13494","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:878976dd53a92cf3bc0fa4f0e43d5a37e641a2d99d84536723c4c6e0b27fabfb","sha256:b36b9ce4bea49f6b6d561eb9575a6c94a67ddfc3b799221519493643fdfa145e"],"state_sha256":"6554e08698eff374b5a996386352162f206c1e95fa26129eea1267640b79c192"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0NOn32ySha5VCzvd7rABE9EyBrPttVGR3gc1xOI1jBIYc4nKjwzPgcc0F7JpECQ+RgcnnSF5LeDpZ2ODQ4tRAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-13T08:41:52.210522Z","bundle_sha256":"a0aea0040adcde860f1323365068396e142f3bf38c993470084332c359b2abff"}}