{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:5YIQT3TAWWWGABPRZQQGYYCDSH","short_pith_number":"pith:5YIQT3TA","canonical_record":{"source":{"id":"2311.09096","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-11-15T16:42:29Z","cross_cats_sorted":[],"title_canon_sha256":"d852c4db4eec1c7c5b9f03ebf5a17e3d7bf3e847a0a7bc81e56a42322ba9a920","abstract_canon_sha256":"a2e1ec2fd5f16d0fd5f9d058b0e293624423016c600c105c04f08731c1c544bc"},"schema_version":"1.0"},"canonical_sha256":"ee1109ee60b5ac6005f1cc206c604391d1fee4329eec7adc76f7f4704aeac748","source":{"kind":"arxiv","id":"2311.09096","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2311.09096","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"arxiv_version","alias_value":"2311.09096v2","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.09096","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"pith_short_12","alias_value":"5YIQT3TAWWWG","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"pith_short_16","alias_value":"5YIQT3TAWWWGABPR","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"pith_short_8","alias_value":"5YIQT3TA","created_at":"2026-07-05T08:30:30Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:5YIQT3TAWWWGABPRZQQGYYCDSH","target":"record","payload":{"canonical_record":{"source":{"id":"2311.09096","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-11-15T16:42:29Z","cross_cats_sorted":[],"title_canon_sha256":"d852c4db4eec1c7c5b9f03ebf5a17e3d7bf3e847a0a7bc81e56a42322ba9a920","abstract_canon_sha256":"a2e1ec2fd5f16d0fd5f9d058b0e293624423016c600c105c04f08731c1c544bc"},"schema_version":"1.0"},"canonical_sha256":"ee1109ee60b5ac6005f1cc206c604391d1fee4329eec7adc76f7f4704aeac748","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:30:30.174827Z","signature_b64":"7M3WF0j53Yr+Gv1ubOuHxnFHw0VIIaXeOMVIgti5huJ2xebfsH89ZQHkXVfqagD0kirZCsRAegCRQZNIGz0KDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ee1109ee60b5ac6005f1cc206c604391d1fee4329eec7adc76f7f4704aeac748","last_reissued_at":"2026-07-05T08:30:30.174307Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:30:30.174307Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2311.09096","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":"2026-07-05T08:30:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"H9YcJm9aUnr5Dw0Qnx00vwWW13COrn8+fZMp2TA2BS4c3ohuhmDAbVOMXwTMUFR4xbHznWkh7hSlZZyEms8gDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T15:05:26.485207Z"},"content_sha256":"11b8a0aa4d5ce872c12bc8a4c799ec1936460e247cdaece3baf2da9af9e85379","schema_version":"1.0","event_id":"sha256:11b8a0aa4d5ce872c12bc8a4c799ec1936460e247cdaece3baf2da9af9e85379"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:5YIQT3TAWWWGABPRZQQGYYCDSH","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Defending Large Language Models Against Jailbreaking Attacks Through Goal Prioritization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Fei Mi, Hongning Wang, Junxiao Yang, Minlie Huang, Pei Ke, Zhexin Zhang","submitted_at":"2023-11-15T16:42:29Z","abstract_excerpt":"While significant attention has been dedicated to exploiting weaknesses in LLMs through jailbreaking attacks, there remains a paucity of effort in defending against these attacks. We point out a pivotal factor contributing to the success of jailbreaks: the intrinsic conflict between the goals of being helpful and ensuring safety. Accordingly, we propose to integrate goal prioritization at both training and inference stages to counteract. Implementing goal prioritization during inference substantially diminishes the Attack Success Rate (ASR) of jailbreaking from 66.4% to 3.6% for ChatGPT. And i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.09096","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/2311.09096/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":"2026-07-05T08:30:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Suf5PKNcxIYVFY2CcJ5DIskhaMgowu437Yq6Sw1kocPLHCKblPCl/tuUYrNLgW4MWA3NVjCSewIktJFBESdvAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T15:05:26.485703Z"},"content_sha256":"7d6b44da288fedf5f221dee718bca89a6882761c61afc06d7ae016f86194f54a","schema_version":"1.0","event_id":"sha256:7d6b44da288fedf5f221dee718bca89a6882761c61afc06d7ae016f86194f54a"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/5YIQT3TAWWWGABPRZQQGYYCDSH/bundle.json","state_url":"https://pith.science/pith/5YIQT3TAWWWGABPRZQQGYYCDSH/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/5YIQT3TAWWWGABPRZQQGYYCDSH/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":"2026-08-06T15:05:26Z","links":{"resolver":"https://pith.science/pith/5YIQT3TAWWWGABPRZQQGYYCDSH","bundle":"https://pith.science/pith/5YIQT3TAWWWGABPRZQQGYYCDSH/bundle.json","state":"https://pith.science/pith/5YIQT3TAWWWGABPRZQQGYYCDSH/state.json","well_known_bundle":"https://pith.science/.well-known/pith/5YIQT3TAWWWGABPRZQQGYYCDSH/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:5YIQT3TAWWWGABPRZQQGYYCDSH","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":"a2e1ec2fd5f16d0fd5f9d058b0e293624423016c600c105c04f08731c1c544bc","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-11-15T16:42:29Z","title_canon_sha256":"d852c4db4eec1c7c5b9f03ebf5a17e3d7bf3e847a0a7bc81e56a42322ba9a920"},"schema_version":"1.0","source":{"id":"2311.09096","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2311.09096","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"arxiv_version","alias_value":"2311.09096v2","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.09096","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"pith_short_12","alias_value":"5YIQT3TAWWWG","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"pith_short_16","alias_value":"5YIQT3TAWWWGABPR","created_at":"2026-07-05T08:30:30Z"},{"alias_kind":"pith_short_8","alias_value":"5YIQT3TA","created_at":"2026-07-05T08:30:30Z"}],"graph_snapshots":[{"event_id":"sha256:7d6b44da288fedf5f221dee718bca89a6882761c61afc06d7ae016f86194f54a","target":"graph","created_at":"2026-07-05T08:30:30Z","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/2311.09096/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"While significant attention has been dedicated to exploiting weaknesses in LLMs through jailbreaking attacks, there remains a paucity of effort in defending against these attacks. We point out a pivotal factor contributing to the success of jailbreaks: the intrinsic conflict between the goals of being helpful and ensuring safety. Accordingly, we propose to integrate goal prioritization at both training and inference stages to counteract. Implementing goal prioritization during inference substantially diminishes the Attack Success Rate (ASR) of jailbreaking from 66.4% to 3.6% for ChatGPT. And i","authors_text":"Fei Mi, Hongning Wang, Junxiao Yang, Minlie Huang, Pei Ke, Zhexin Zhang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-11-15T16:42:29Z","title":"Defending Large Language Models Against Jailbreaking Attacks Through Goal Prioritization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.09096","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":"sha256:11b8a0aa4d5ce872c12bc8a4c799ec1936460e247cdaece3baf2da9af9e85379","target":"record","created_at":"2026-07-05T08:30:30Z","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":"a2e1ec2fd5f16d0fd5f9d058b0e293624423016c600c105c04f08731c1c544bc","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-11-15T16:42:29Z","title_canon_sha256":"d852c4db4eec1c7c5b9f03ebf5a17e3d7bf3e847a0a7bc81e56a42322ba9a920"},"schema_version":"1.0","source":{"id":"2311.09096","kind":"arxiv","version":2}},"canonical_sha256":"ee1109ee60b5ac6005f1cc206c604391d1fee4329eec7adc76f7f4704aeac748","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ee1109ee60b5ac6005f1cc206c604391d1fee4329eec7adc76f7f4704aeac748","first_computed_at":"2026-07-05T08:30:30.174307Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:30:30.174307Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"7M3WF0j53Yr+Gv1ubOuHxnFHw0VIIaXeOMVIgti5huJ2xebfsH89ZQHkXVfqagD0kirZCsRAegCRQZNIGz0KDA==","signature_status":"signed_v1","signed_at":"2026-07-05T08:30:30.174827Z","signed_message":"canonical_sha256_bytes"},"source_id":"2311.09096","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:11b8a0aa4d5ce872c12bc8a4c799ec1936460e247cdaece3baf2da9af9e85379","sha256:7d6b44da288fedf5f221dee718bca89a6882761c61afc06d7ae016f86194f54a"],"state_sha256":"9c7c905245ff52883f01111a3c753ca190cc63577117ddd5048cc82f2af43f93"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"v3FXBfV6cd63FYHdj7Sa0EoOlT7/aV0nGMcnYNojpk42QAPo/uerOZsC+lIlG4ha90X9DDVExDPAnRrYpDYkCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T15:05:26.490186Z","bundle_sha256":"d81b22878d57969d41a8c9fa2b93098d4c7d44a828384957878bcfd3e2802c1d"}}