{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:XTYQAXSJRWYVCWAZLX7HQTHUNH","short_pith_number":"pith:XTYQAXSJ","canonical_record":{"source":{"id":"2406.10794","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-16T03:38:48Z","cross_cats_sorted":[],"title_canon_sha256":"260e3f0dae6bd6d719abb2f30c3f2fa89276a646e9d1814eca75cc82d93037e2","abstract_canon_sha256":"e1a1f8ab15da914d8ee154c5427423fa87a75fd235925f30921454d387b2227f"},"schema_version":"1.0"},"canonical_sha256":"bcf1005e498db15158195dfe784cf469c2e52a368b8abd9e8a22e48f9bd97c82","source":{"kind":"arxiv","id":"2406.10794","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.10794","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"arxiv_version","alias_value":"2406.10794v3","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.10794","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"pith_short_12","alias_value":"XTYQAXSJRWYV","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"pith_short_16","alias_value":"XTYQAXSJRWYVCWAZ","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"pith_short_8","alias_value":"XTYQAXSJ","created_at":"2026-07-05T09:43:30Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:XTYQAXSJRWYVCWAZLX7HQTHUNH","target":"record","payload":{"canonical_record":{"source":{"id":"2406.10794","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-16T03:38:48Z","cross_cats_sorted":[],"title_canon_sha256":"260e3f0dae6bd6d719abb2f30c3f2fa89276a646e9d1814eca75cc82d93037e2","abstract_canon_sha256":"e1a1f8ab15da914d8ee154c5427423fa87a75fd235925f30921454d387b2227f"},"schema_version":"1.0"},"canonical_sha256":"bcf1005e498db15158195dfe784cf469c2e52a368b8abd9e8a22e48f9bd97c82","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:43:30.320455Z","signature_b64":"OkswiHIbSwdsqcD7FmYbH3KdBJbiqHmWB8KDSquKVzuZJIm7xfRbPfYVMxqTUrIWvI61z9i+ANSZMgtVrDulCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bcf1005e498db15158195dfe784cf469c2e52a368b8abd9e8a22e48f9bd97c82","last_reissued_at":"2026-07-05T09:43:30.319882Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:43:30.319882Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.10794","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":"2026-07-05T09:43:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LAzvh1SgcIGB9z2DT8fQBtHBxnw7ShKxCqx6qAlQQLzyuFVMCKhZfv9EG8OTbIwl2wbioggzx1rrNxY3DMZfAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T09:50:49.006324Z"},"content_sha256":"eef50971074f7a689879d84d35e5e1c1b1f86ad2dc6e54a4cf014ea19c1c4e93","schema_version":"1.0","event_id":"sha256:eef50971074f7a689879d84d35e5e1c1b1f86ad2dc6e54a4cf014ea19c1c4e93"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:XTYQAXSJRWYVCWAZLX7HQTHUNH","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Towards Understanding Jailbreak Attacks in LLMs: A Representation Space Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Han Xu, Hui Liu, Jiliang Tang, Makoto Yamada, Pengfei He, Yue Xing, Yuping Lin","submitted_at":"2024-06-16T03:38:48Z","abstract_excerpt":"Large language models (LLMs) are susceptible to a type of attack known as jailbreaking, which misleads LLMs to output harmful contents. Although there are diverse jailbreak attack strategies, there is no unified understanding on why some methods succeed and others fail. This paper explores the behavior of harmful and harmless prompts in the LLM's representation space to investigate the intrinsic properties of successful jailbreak attacks. We hypothesize that successful attacks share some similar properties: They are effective in moving the representation of the harmful prompt towards the direc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.10794","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/2406.10794/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":"2026-07-05T09:43:30Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UhD5QPliySizI6ODKXVFFZLBzYJhYsCIqHbnKCM+zcBjY+nt3+0+uvRK0N7eYDgrxbSKtg+DgRpsx9iMeJVAAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T09:50:49.006882Z"},"content_sha256":"ae9772e7951d97b79a3b5bc45f8fc6b44cc9371d433814ed196a645da0654b69","schema_version":"1.0","event_id":"sha256:ae9772e7951d97b79a3b5bc45f8fc6b44cc9371d433814ed196a645da0654b69"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XTYQAXSJRWYVCWAZLX7HQTHUNH/bundle.json","state_url":"https://pith.science/pith/XTYQAXSJRWYVCWAZLX7HQTHUNH/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XTYQAXSJRWYVCWAZLX7HQTHUNH/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":"2026-08-06T09:50:49Z","links":{"resolver":"https://pith.science/pith/XTYQAXSJRWYVCWAZLX7HQTHUNH","bundle":"https://pith.science/pith/XTYQAXSJRWYVCWAZLX7HQTHUNH/bundle.json","state":"https://pith.science/pith/XTYQAXSJRWYVCWAZLX7HQTHUNH/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XTYQAXSJRWYVCWAZLX7HQTHUNH/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:XTYQAXSJRWYVCWAZLX7HQTHUNH","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":"e1a1f8ab15da914d8ee154c5427423fa87a75fd235925f30921454d387b2227f","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-16T03:38:48Z","title_canon_sha256":"260e3f0dae6bd6d719abb2f30c3f2fa89276a646e9d1814eca75cc82d93037e2"},"schema_version":"1.0","source":{"id":"2406.10794","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.10794","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"arxiv_version","alias_value":"2406.10794v3","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.10794","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"pith_short_12","alias_value":"XTYQAXSJRWYV","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"pith_short_16","alias_value":"XTYQAXSJRWYVCWAZ","created_at":"2026-07-05T09:43:30Z"},{"alias_kind":"pith_short_8","alias_value":"XTYQAXSJ","created_at":"2026-07-05T09:43:30Z"}],"graph_snapshots":[{"event_id":"sha256:ae9772e7951d97b79a3b5bc45f8fc6b44cc9371d433814ed196a645da0654b69","target":"graph","created_at":"2026-07-05T09:43:30Z","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/2406.10794/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are susceptible to a type of attack known as jailbreaking, which misleads LLMs to output harmful contents. Although there are diverse jailbreak attack strategies, there is no unified understanding on why some methods succeed and others fail. This paper explores the behavior of harmful and harmless prompts in the LLM's representation space to investigate the intrinsic properties of successful jailbreak attacks. We hypothesize that successful attacks share some similar properties: They are effective in moving the representation of the harmful prompt towards the direc","authors_text":"Han Xu, Hui Liu, Jiliang Tang, Makoto Yamada, Pengfei He, Yue Xing, Yuping Lin","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-16T03:38:48Z","title":"Towards Understanding Jailbreak Attacks in LLMs: A Representation Space Analysis"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.10794","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:eef50971074f7a689879d84d35e5e1c1b1f86ad2dc6e54a4cf014ea19c1c4e93","target":"record","created_at":"2026-07-05T09:43:30Z","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":"e1a1f8ab15da914d8ee154c5427423fa87a75fd235925f30921454d387b2227f","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-16T03:38:48Z","title_canon_sha256":"260e3f0dae6bd6d719abb2f30c3f2fa89276a646e9d1814eca75cc82d93037e2"},"schema_version":"1.0","source":{"id":"2406.10794","kind":"arxiv","version":3}},"canonical_sha256":"bcf1005e498db15158195dfe784cf469c2e52a368b8abd9e8a22e48f9bd97c82","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bcf1005e498db15158195dfe784cf469c2e52a368b8abd9e8a22e48f9bd97c82","first_computed_at":"2026-07-05T09:43:30.319882Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:43:30.319882Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"OkswiHIbSwdsqcD7FmYbH3KdBJbiqHmWB8KDSquKVzuZJIm7xfRbPfYVMxqTUrIWvI61z9i+ANSZMgtVrDulCQ==","signature_status":"signed_v1","signed_at":"2026-07-05T09:43:30.320455Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.10794","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:eef50971074f7a689879d84d35e5e1c1b1f86ad2dc6e54a4cf014ea19c1c4e93","sha256:ae9772e7951d97b79a3b5bc45f8fc6b44cc9371d433814ed196a645da0654b69"],"state_sha256":"6bdefaecf3fe6146cb6e8b9d8d19adae408abeefff7b62287f02eedf05f219a5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zI8I7AVlAQbNiJPKrWqA3zSEV3PeGrIOwM15x3qREpByN0poUiDqdikba1o7LJ0TEDgoyfvtMZl9PCNhDNk7CQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T09:50:49.010529Z","bundle_sha256":"b191bf973f95d8ad0aa90ce49a7b7011b02d8b5d1fa752d5ecbbca74fedc0183"}}