{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:YH7RP5Y2PSHCOUFLGJXJDUEA34","short_pith_number":"pith:YH7RP5Y2","canonical_record":{"source":{"id":"2403.00867","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-03-01T03:29:54Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"b6a7d4d270430dfaf03c230930c90290514baa6ffbe3a481c1aa0ab70f69710e","abstract_canon_sha256":"5a561f4a8aed206885ba439b8e8473ccc39884546a4ccdf5fe5273e230210438"},"schema_version":"1.0"},"canonical_sha256":"c1ff17f71a7c8e2750ab326e91d080df10c431860ddd90826b64f79213beab59","source":{"kind":"arxiv","id":"2403.00867","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.00867","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"arxiv_version","alias_value":"2403.00867v3","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.00867","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"pith_short_12","alias_value":"YH7RP5Y2PSHC","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"pith_short_16","alias_value":"YH7RP5Y2PSHCOUFL","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"pith_short_8","alias_value":"YH7RP5Y2","created_at":"2026-07-05T09:32:31Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:YH7RP5Y2PSHCOUFLGJXJDUEA34","target":"record","payload":{"canonical_record":{"source":{"id":"2403.00867","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-03-01T03:29:54Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"b6a7d4d270430dfaf03c230930c90290514baa6ffbe3a481c1aa0ab70f69710e","abstract_canon_sha256":"5a561f4a8aed206885ba439b8e8473ccc39884546a4ccdf5fe5273e230210438"},"schema_version":"1.0"},"canonical_sha256":"c1ff17f71a7c8e2750ab326e91d080df10c431860ddd90826b64f79213beab59","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:32:31.534751Z","signature_b64":"4PjCdvlq/eWrXAnoQ4EMrpckgWhoQW0O36Paz4jqiEHbFvRN1lXOB5zxuEwrcXJfeSvYf3tDWWlsMFq86kyPBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c1ff17f71a7c8e2750ab326e91d080df10c431860ddd90826b64f79213beab59","last_reissued_at":"2026-07-05T09:32:31.534236Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:32:31.534236Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2403.00867","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:32:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xueRL5s4ZJZ0pkv0jAdqjV8/BmQepphgnWzNUkD191k14215TePdQLwm9QZmxqCE5cxwimkGGfxFQuqsZC/EBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T12:28:24.425733Z"},"content_sha256":"f32a2ed3e010ca7c53245024367c39fbcc086f54ef95745162d1e5ee7659c6df","schema_version":"1.0","event_id":"sha256:f32a2ed3e010ca7c53245024367c39fbcc086f54ef95745162d1e5ee7659c6df"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:YH7RP5Y2PSHCOUFLGJXJDUEA34","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Gradient Cuff: Detecting Jailbreak Attacks on Large Language Models by Exploring Refusal Loss Landscapes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CR","authors_text":"Pin-Yu Chen, Tsung-Yi Ho, Xiaomeng Hu","submitted_at":"2024-03-01T03:29:54Z","abstract_excerpt":"Large Language Models (LLMs) are becoming a prominent generative AI tool, where the user enters a query and the LLM generates an answer. To reduce harm and misuse, efforts have been made to align these LLMs to human values using advanced training techniques such as Reinforcement Learning from Human Feedback (RLHF). However, recent studies have highlighted the vulnerability of LLMs to adversarial jailbreak attempts aiming at subverting the embedded safety guardrails. To address this challenge, this paper defines and investigates the Refusal Loss of LLMs and then proposes a method called Gradien"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.00867","kind":"arxiv","version":3},"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/2403.00867/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":"2026-07-05T09:32:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"bS5jqdGxgsXthsacHlMr2Z5fmUNmdA+yAnMtNDIclBqsZrJbsTuTFjlFNkmhrRfjpGA9imNx+T37M7UlVQLODg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T12:28:24.426250Z"},"content_sha256":"284f3a32e29d40d6246ada14280e49bc11ed9531d698960387d021442afe3565","schema_version":"1.0","event_id":"sha256:284f3a32e29d40d6246ada14280e49bc11ed9531d698960387d021442afe3565"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/YH7RP5Y2PSHCOUFLGJXJDUEA34/bundle.json","state_url":"https://pith.science/pith/YH7RP5Y2PSHCOUFLGJXJDUEA34/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/YH7RP5Y2PSHCOUFLGJXJDUEA34/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":"2026-08-09T12:28:24Z","links":{"resolver":"https://pith.science/pith/YH7RP5Y2PSHCOUFLGJXJDUEA34","bundle":"https://pith.science/pith/YH7RP5Y2PSHCOUFLGJXJDUEA34/bundle.json","state":"https://pith.science/pith/YH7RP5Y2PSHCOUFLGJXJDUEA34/state.json","well_known_bundle":"https://pith.science/.well-known/pith/YH7RP5Y2PSHCOUFLGJXJDUEA34/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:YH7RP5Y2PSHCOUFLGJXJDUEA34","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":"5a561f4a8aed206885ba439b8e8473ccc39884546a4ccdf5fe5273e230210438","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-03-01T03:29:54Z","title_canon_sha256":"b6a7d4d270430dfaf03c230930c90290514baa6ffbe3a481c1aa0ab70f69710e"},"schema_version":"1.0","source":{"id":"2403.00867","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.00867","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"arxiv_version","alias_value":"2403.00867v3","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.00867","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"pith_short_12","alias_value":"YH7RP5Y2PSHC","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"pith_short_16","alias_value":"YH7RP5Y2PSHCOUFL","created_at":"2026-07-05T09:32:31Z"},{"alias_kind":"pith_short_8","alias_value":"YH7RP5Y2","created_at":"2026-07-05T09:32:31Z"}],"graph_snapshots":[{"event_id":"sha256:284f3a32e29d40d6246ada14280e49bc11ed9531d698960387d021442afe3565","target":"graph","created_at":"2026-07-05T09:32:31Z","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/2403.00867/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) are becoming a prominent generative AI tool, where the user enters a query and the LLM generates an answer. To reduce harm and misuse, efforts have been made to align these LLMs to human values using advanced training techniques such as Reinforcement Learning from Human Feedback (RLHF). However, recent studies have highlighted the vulnerability of LLMs to adversarial jailbreak attempts aiming at subverting the embedded safety guardrails. To address this challenge, this paper defines and investigates the Refusal Loss of LLMs and then proposes a method called Gradien","authors_text":"Pin-Yu Chen, Tsung-Yi Ho, Xiaomeng Hu","cross_cats":["cs.AI","cs.CL","cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-03-01T03:29:54Z","title":"Gradient Cuff: Detecting Jailbreak Attacks on Large Language Models by Exploring Refusal Loss Landscapes"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.00867","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:f32a2ed3e010ca7c53245024367c39fbcc086f54ef95745162d1e5ee7659c6df","target":"record","created_at":"2026-07-05T09:32:31Z","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":"5a561f4a8aed206885ba439b8e8473ccc39884546a4ccdf5fe5273e230210438","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-03-01T03:29:54Z","title_canon_sha256":"b6a7d4d270430dfaf03c230930c90290514baa6ffbe3a481c1aa0ab70f69710e"},"schema_version":"1.0","source":{"id":"2403.00867","kind":"arxiv","version":3}},"canonical_sha256":"c1ff17f71a7c8e2750ab326e91d080df10c431860ddd90826b64f79213beab59","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c1ff17f71a7c8e2750ab326e91d080df10c431860ddd90826b64f79213beab59","first_computed_at":"2026-07-05T09:32:31.534236Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:32:31.534236Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"4PjCdvlq/eWrXAnoQ4EMrpckgWhoQW0O36Paz4jqiEHbFvRN1lXOB5zxuEwrcXJfeSvYf3tDWWlsMFq86kyPBA==","signature_status":"signed_v1","signed_at":"2026-07-05T09:32:31.534751Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.00867","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f32a2ed3e010ca7c53245024367c39fbcc086f54ef95745162d1e5ee7659c6df","sha256:284f3a32e29d40d6246ada14280e49bc11ed9531d698960387d021442afe3565"],"state_sha256":"00dce061a1bdf6bd1bff5eb3f89b5f21bffc80ac87c04004d1a83570b0f1ddb5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AQyH3YXNjscBdv0kylUmglRl31ql7Z+zETXjUKXGScrPCRNAQGU7miJvQlZlZNnwmxR3Hr/uU2e9YbOMW4RpBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T12:28:24.431694Z","bundle_sha256":"ad880d543879efe00b84b3c4c83412498b44f44582a6287e3941b707c75687d9"}}