{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:C3YYBOA3DGB2WV4D6YEZA5BDGZ","short_pith_number":"pith:C3YYBOA3","canonical_record":{"source":{"id":"2308.07308","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-14T17:54:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8e0c02c99cc886cf777fa9f585be608b79729ead0ad64aa8e95e86c56e7c4bd5","abstract_canon_sha256":"142e4bba90b96274a4e0c67453f8f86499d3000a8c5b52b2cae1e272c3c65ca1"},"schema_version":"1.0"},"canonical_sha256":"16f180b81b1983ab5783f6099074233675efe02f18032bad10d9c3818c770792","source":{"kind":"arxiv","id":"2308.07308","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2308.07308","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"arxiv_version","alias_value":"2308.07308v4","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.07308","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"pith_short_12","alias_value":"C3YYBOA3DGB2","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"pith_short_16","alias_value":"C3YYBOA3DGB2WV4D","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"pith_short_8","alias_value":"C3YYBOA3","created_at":"2026-07-05T08:14:39Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:C3YYBOA3DGB2WV4D6YEZA5BDGZ","target":"record","payload":{"canonical_record":{"source":{"id":"2308.07308","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-14T17:54:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8e0c02c99cc886cf777fa9f585be608b79729ead0ad64aa8e95e86c56e7c4bd5","abstract_canon_sha256":"142e4bba90b96274a4e0c67453f8f86499d3000a8c5b52b2cae1e272c3c65ca1"},"schema_version":"1.0"},"canonical_sha256":"16f180b81b1983ab5783f6099074233675efe02f18032bad10d9c3818c770792","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:14:39.358795Z","signature_b64":"bLDNTrWFyGcmrpxev0D5tdVfq+XjVdHVwTbI7Db8l5bVbx+dyaU8oNbjd8y2AYqS4qFsevDXf+KQNCKJuxGQBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"16f180b81b1983ab5783f6099074233675efe02f18032bad10d9c3818c770792","last_reissued_at":"2026-07-05T08:14:39.358325Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:14:39.358325Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2308.07308","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:14:39Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/0JoK4T2NnlBHzlEiCZ37JQX68hbowXIvg5no6cQXo+KYFv+EIJ/eWOkKkNzqo+yb0zNFVhRduXua6uaG6ZoBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-18T16:22:41.943715Z"},"content_sha256":"b56b4a43f6cb7f10a981d0d8ca1ca54ddcf097bc07e116f2372aad373cb745c8","schema_version":"1.0","event_id":"sha256:b56b4a43f6cb7f10a981d0d8ca1ca54ddcf097bc07e116f2372aad373cb745c8"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:C3YYBOA3DGB2WV4D6YEZA5BDGZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"LLM Self Defense: By Self Examination, LLMs Know They Are Being Tricked","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Alec Helbling, Cory Cornelius, Duen Horng Chau, Mansi Phute, Matthew Hull, Sebastian Szyller, ShengYun Peng","submitted_at":"2023-08-14T17:54:10Z","abstract_excerpt":"Large language models (LLMs) are popular for high-quality text generation but can produce harmful content, even when aligned with human values through reinforcement learning. Adversarial prompts can bypass their safety measures. We propose LLM Self Defense, a simple approach to defend against these attacks by having an LLM screen the induced responses. Our method does not require any fine-tuning, input preprocessing, or iterative output generation. Instead, we incorporate the generated content into a pre-defined prompt and employ another instance of an LLM to analyze the text and predict wheth"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.07308","kind":"arxiv","version":4},"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/2308.07308/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":"2026-07-05T08:14:39Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+NWe3YPweDluIjARWBZE++T0kS2PH/9AqdbxsDNyC0i8QNQt/xzKw8JcRkT2OWEpncAqaSxCYkbI5Zbb2G1DBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-18T16:22:41.944654Z"},"content_sha256":"6030e9d10433bd255101fec9c0eb20747fb362ecc9603ab775c3fc14d02b5dcf","schema_version":"1.0","event_id":"sha256:6030e9d10433bd255101fec9c0eb20747fb362ecc9603ab775c3fc14d02b5dcf"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/C3YYBOA3DGB2WV4D6YEZA5BDGZ/bundle.json","state_url":"https://pith.science/pith/C3YYBOA3DGB2WV4D6YEZA5BDGZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/C3YYBOA3DGB2WV4D6YEZA5BDGZ/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":"2026-08-18T16:22:41Z","links":{"resolver":"https://pith.science/pith/C3YYBOA3DGB2WV4D6YEZA5BDGZ","bundle":"https://pith.science/pith/C3YYBOA3DGB2WV4D6YEZA5BDGZ/bundle.json","state":"https://pith.science/pith/C3YYBOA3DGB2WV4D6YEZA5BDGZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/C3YYBOA3DGB2WV4D6YEZA5BDGZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:C3YYBOA3DGB2WV4D6YEZA5BDGZ","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":"142e4bba90b96274a4e0c67453f8f86499d3000a8c5b52b2cae1e272c3c65ca1","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-14T17:54:10Z","title_canon_sha256":"8e0c02c99cc886cf777fa9f585be608b79729ead0ad64aa8e95e86c56e7c4bd5"},"schema_version":"1.0","source":{"id":"2308.07308","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2308.07308","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"arxiv_version","alias_value":"2308.07308v4","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.07308","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"pith_short_12","alias_value":"C3YYBOA3DGB2","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"pith_short_16","alias_value":"C3YYBOA3DGB2WV4D","created_at":"2026-07-05T08:14:39Z"},{"alias_kind":"pith_short_8","alias_value":"C3YYBOA3","created_at":"2026-07-05T08:14:39Z"}],"graph_snapshots":[{"event_id":"sha256:6030e9d10433bd255101fec9c0eb20747fb362ecc9603ab775c3fc14d02b5dcf","target":"graph","created_at":"2026-07-05T08:14:39Z","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/2308.07308/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are popular for high-quality text generation but can produce harmful content, even when aligned with human values through reinforcement learning. Adversarial prompts can bypass their safety measures. We propose LLM Self Defense, a simple approach to defend against these attacks by having an LLM screen the induced responses. Our method does not require any fine-tuning, input preprocessing, or iterative output generation. Instead, we incorporate the generated content into a pre-defined prompt and employ another instance of an LLM to analyze the text and predict wheth","authors_text":"Alec Helbling, Cory Cornelius, Duen Horng Chau, Mansi Phute, Matthew Hull, Sebastian Szyller, ShengYun Peng","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-14T17:54:10Z","title":"LLM Self Defense: By Self Examination, LLMs Know They Are Being Tricked"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.07308","kind":"arxiv","version":4},"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:b56b4a43f6cb7f10a981d0d8ca1ca54ddcf097bc07e116f2372aad373cb745c8","target":"record","created_at":"2026-07-05T08:14:39Z","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":"142e4bba90b96274a4e0c67453f8f86499d3000a8c5b52b2cae1e272c3c65ca1","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-14T17:54:10Z","title_canon_sha256":"8e0c02c99cc886cf777fa9f585be608b79729ead0ad64aa8e95e86c56e7c4bd5"},"schema_version":"1.0","source":{"id":"2308.07308","kind":"arxiv","version":4}},"canonical_sha256":"16f180b81b1983ab5783f6099074233675efe02f18032bad10d9c3818c770792","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"16f180b81b1983ab5783f6099074233675efe02f18032bad10d9c3818c770792","first_computed_at":"2026-07-05T08:14:39.358325Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:14:39.358325Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"bLDNTrWFyGcmrpxev0D5tdVfq+XjVdHVwTbI7Db8l5bVbx+dyaU8oNbjd8y2AYqS4qFsevDXf+KQNCKJuxGQBw==","signature_status":"signed_v1","signed_at":"2026-07-05T08:14:39.358795Z","signed_message":"canonical_sha256_bytes"},"source_id":"2308.07308","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b56b4a43f6cb7f10a981d0d8ca1ca54ddcf097bc07e116f2372aad373cb745c8","sha256:6030e9d10433bd255101fec9c0eb20747fb362ecc9603ab775c3fc14d02b5dcf"],"state_sha256":"f54cf6f601bb0fc0a2202e720c6a127f261e382a65b9548aa4bd3153a6e93ea0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"IAMJMSYZQCd0yKTC7f7GVmFnJERk661RqbongccM3IT7oML901iXsKBVU0x746g8M8LlyimA60/5uM8H4GNvDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-18T16:22:41.950271Z","bundle_sha256":"300de03dacb83de8e82e3ad93c9592203f126fe6d202c6504adf0991123f79ce"}}