{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:RGEI5KAK5JDYJ2YKPWPY5S4W5F","short_pith_number":"pith:RGEI5KAK","canonical_record":{"source":{"id":"2409.03788","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-08-31T06:50:07Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"fb27f1132a2abc35a168a4bf26d1206ec1e69aa03dbca70c0439fd772d6b5cbf","abstract_canon_sha256":"90436e3f41c0f27a21fd51aa94fb33905731cd48452214b3bcc104502491795c"},"schema_version":"1.0"},"canonical_sha256":"89888ea80aea4784eb0a7d9f8ecb96e97f046ec77a24d47fecc70dcbd2916636","source":{"kind":"arxiv","id":"2409.03788","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2409.03788","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"arxiv_version","alias_value":"2409.03788v2","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.03788","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"pith_short_12","alias_value":"RGEI5KAK5JDY","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"pith_short_16","alias_value":"RGEI5KAK5JDYJ2YK","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"pith_short_8","alias_value":"RGEI5KAK","created_at":"2026-07-05T11:18:13Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:RGEI5KAK5JDYJ2YKPWPY5S4W5F","target":"record","payload":{"canonical_record":{"source":{"id":"2409.03788","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-08-31T06:50:07Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"fb27f1132a2abc35a168a4bf26d1206ec1e69aa03dbca70c0439fd772d6b5cbf","abstract_canon_sha256":"90436e3f41c0f27a21fd51aa94fb33905731cd48452214b3bcc104502491795c"},"schema_version":"1.0"},"canonical_sha256":"89888ea80aea4784eb0a7d9f8ecb96e97f046ec77a24d47fecc70dcbd2916636","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:18:13.969531Z","signature_b64":"e28bt4aNi/yQosEEhGuZapw9qelFiuvbiZq4nbU6Q0LiL/dhgf60A33ZefFgp8ul7ua4w1xE5yuVzoBGYxQVDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"89888ea80aea4784eb0a7d9f8ecb96e97f046ec77a24d47fecc70dcbd2916636","last_reissued_at":"2026-07-05T11:18:13.969063Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:18:13.969063Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2409.03788","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":"2026-07-05T11:18:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Tl0hfjva01bs5SuMZG0NBR73kJIXFfs22BIyMlqVI1FH5h8xa0nStrYLtbxfIfSbatl8FqPWN8X37ef/3kmJCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T03:30:23.455220Z"},"content_sha256":"5362fac39de711eb61fec7a9552cf793bbdefb4ca965584952a4c562d399873c","schema_version":"1.0","event_id":"sha256:5362fac39de711eb61fec7a9552cf793bbdefb4ca965584952a4c562d399873c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:RGEI5KAK5JDYJ2YKPWPY5S4W5F","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"HSF: Defending against Jailbreak Attacks with Hidden State Filtering","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CR","authors_text":"Cheng Qian, Hainan Zhang, Lei Sha, Zhiming Zheng","submitted_at":"2024-08-31T06:50:07Z","abstract_excerpt":"With the growing deployment of LLMs in daily applications like chatbots and content generation, efforts to ensure outputs align with human values and avoid harmful content have intensified. However, increasingly sophisticated jailbreak attacks threaten this alignment, aiming to induce unsafe outputs. Current defense efforts either focus on prompt rewriting or detection, which are limited in effectiveness due to the various design of jailbreak prompts, or on output control and detection, which are computationally expensive as they require LLM inference. Therefore, designing a pre-inference defe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.03788","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/2409.03788/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":"2026-07-05T11:18:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9eht4mI8T+ilU/pw42GuOC5f2wHbko9DZaP9TnTdK//qEnjNjkCOd7sqmN3qdXvE22vj5E+/sxRTvxb5KPYaBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T03:30:23.455748Z"},"content_sha256":"23be9ade1aef4433a99df6abaa933bcf4135d8f1cfcdabedc8b39f74ac589cb1","schema_version":"1.0","event_id":"sha256:23be9ade1aef4433a99df6abaa933bcf4135d8f1cfcdabedc8b39f74ac589cb1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/RGEI5KAK5JDYJ2YKPWPY5S4W5F/bundle.json","state_url":"https://pith.science/pith/RGEI5KAK5JDYJ2YKPWPY5S4W5F/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/RGEI5KAK5JDYJ2YKPWPY5S4W5F/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":"2026-08-17T03:30:23Z","links":{"resolver":"https://pith.science/pith/RGEI5KAK5JDYJ2YKPWPY5S4W5F","bundle":"https://pith.science/pith/RGEI5KAK5JDYJ2YKPWPY5S4W5F/bundle.json","state":"https://pith.science/pith/RGEI5KAK5JDYJ2YKPWPY5S4W5F/state.json","well_known_bundle":"https://pith.science/.well-known/pith/RGEI5KAK5JDYJ2YKPWPY5S4W5F/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:RGEI5KAK5JDYJ2YKPWPY5S4W5F","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":"90436e3f41c0f27a21fd51aa94fb33905731cd48452214b3bcc104502491795c","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-08-31T06:50:07Z","title_canon_sha256":"fb27f1132a2abc35a168a4bf26d1206ec1e69aa03dbca70c0439fd772d6b5cbf"},"schema_version":"1.0","source":{"id":"2409.03788","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2409.03788","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"arxiv_version","alias_value":"2409.03788v2","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.03788","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"pith_short_12","alias_value":"RGEI5KAK5JDY","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"pith_short_16","alias_value":"RGEI5KAK5JDYJ2YK","created_at":"2026-07-05T11:18:13Z"},{"alias_kind":"pith_short_8","alias_value":"RGEI5KAK","created_at":"2026-07-05T11:18:13Z"}],"graph_snapshots":[{"event_id":"sha256:23be9ade1aef4433a99df6abaa933bcf4135d8f1cfcdabedc8b39f74ac589cb1","target":"graph","created_at":"2026-07-05T11:18:13Z","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/2409.03788/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"With the growing deployment of LLMs in daily applications like chatbots and content generation, efforts to ensure outputs align with human values and avoid harmful content have intensified. However, increasingly sophisticated jailbreak attacks threaten this alignment, aiming to induce unsafe outputs. Current defense efforts either focus on prompt rewriting or detection, which are limited in effectiveness due to the various design of jailbreak prompts, or on output control and detection, which are computationally expensive as they require LLM inference. Therefore, designing a pre-inference defe","authors_text":"Cheng Qian, Hainan Zhang, Lei Sha, Zhiming Zheng","cross_cats":["cs.AI","cs.CL","cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-08-31T06:50:07Z","title":"HSF: Defending against Jailbreak Attacks with Hidden State Filtering"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.03788","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":"sha256:5362fac39de711eb61fec7a9552cf793bbdefb4ca965584952a4c562d399873c","target":"record","created_at":"2026-07-05T11:18:13Z","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":"90436e3f41c0f27a21fd51aa94fb33905731cd48452214b3bcc104502491795c","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-08-31T06:50:07Z","title_canon_sha256":"fb27f1132a2abc35a168a4bf26d1206ec1e69aa03dbca70c0439fd772d6b5cbf"},"schema_version":"1.0","source":{"id":"2409.03788","kind":"arxiv","version":2}},"canonical_sha256":"89888ea80aea4784eb0a7d9f8ecb96e97f046ec77a24d47fecc70dcbd2916636","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"89888ea80aea4784eb0a7d9f8ecb96e97f046ec77a24d47fecc70dcbd2916636","first_computed_at":"2026-07-05T11:18:13.969063Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:18:13.969063Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"e28bt4aNi/yQosEEhGuZapw9qelFiuvbiZq4nbU6Q0LiL/dhgf60A33ZefFgp8ul7ua4w1xE5yuVzoBGYxQVDQ==","signature_status":"signed_v1","signed_at":"2026-07-05T11:18:13.969531Z","signed_message":"canonical_sha256_bytes"},"source_id":"2409.03788","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5362fac39de711eb61fec7a9552cf793bbdefb4ca965584952a4c562d399873c","sha256:23be9ade1aef4433a99df6abaa933bcf4135d8f1cfcdabedc8b39f74ac589cb1"],"state_sha256":"a706e5f4afd5b4563b788b17f5c53d2842cb99431d92c26ce78c72e411c8d3f0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Pyt93sYy7DPeddsMhHF8R9IejYskyYd43vCTBNdC2aBFUH9SMAy29iEULEG+LQ31lMMdF6Ak7cAcKvWlCpDiCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T03:30:23.459273Z","bundle_sha256":"7d837a372e477fbe49f6ba19d42d77eaa159f5183a6f96e6c4da4b4cc1bf7f1a"}}