{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:RTGGN4I6CAOIN5IQTZFLUOCXPZ","short_pith_number":"pith:RTGGN4I6","canonical_record":{"source":{"id":"2311.00172","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-31T22:22:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"577bffaf64f07e67ca3d3fefaf5005df2f94c3b38488d66dd4531b3c300c1575","abstract_canon_sha256":"5fd58d9c9ad37bc5fd17e3cc097c452f4eaa20ab0b2cf77b5b71d12e6549572c"},"schema_version":"1.0"},"canonical_sha256":"8ccc66f11e101c86f5109e4aba38577e75945bc02d9c8eb812a63f0e376d8c81","source":{"kind":"arxiv","id":"2311.00172","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2311.00172","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"arxiv_version","alias_value":"2311.00172v1","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.00172","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"pith_short_12","alias_value":"RTGGN4I6CAOI","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"pith_short_16","alias_value":"RTGGN4I6CAOIN5IQ","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"pith_short_8","alias_value":"RTGGN4I6","created_at":"2026-07-05T07:07:43Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:RTGGN4I6CAOIN5IQTZFLUOCXPZ","target":"record","payload":{"canonical_record":{"source":{"id":"2311.00172","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-31T22:22:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"577bffaf64f07e67ca3d3fefaf5005df2f94c3b38488d66dd4531b3c300c1575","abstract_canon_sha256":"5fd58d9c9ad37bc5fd17e3cc097c452f4eaa20ab0b2cf77b5b71d12e6549572c"},"schema_version":"1.0"},"canonical_sha256":"8ccc66f11e101c86f5109e4aba38577e75945bc02d9c8eb812a63f0e376d8c81","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:07:43.213189Z","signature_b64":"2gcxq27ZQziUL9/MoLnRMg0z7BOfoDcAq6nVNJeaRecbike0rMKzJq8PaV6xo/0pFMGEpeCgNi3OgFUsC8M1Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ccc66f11e101c86f5109e4aba38577e75945bc02d9c8eb812a63f0e376d8c81","last_reissued_at":"2026-07-05T07:07:43.212682Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:07:43.212682Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2311.00172","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:07:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oG8jugWBSwaZsE3VvN4zpxvSHBaSg9iRp29Z7+ly1/pKSSwzjTxzrDxbdIiodrYTFWSGS+rWElQBHh71/BVgDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T07:39:12.937743Z"},"content_sha256":"16813928c3febb39de91423b7cb550737d127fa1dc2223ef1be2df7a67dc8b68","schema_version":"1.0","event_id":"sha256:16813928c3febb39de91423b7cb550737d127fa1dc2223ef1be2df7a67dc8b68"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:RTGGN4I6CAOIN5IQTZFLUOCXPZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Robust Safety Classifier for Large Language Models: Adversarial Prompt Shield","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ali Derakhshan, Ian G. Harris, Jinhwa Kim","submitted_at":"2023-10-31T22:22:10Z","abstract_excerpt":"Large Language Models' safety remains a critical concern due to their vulnerability to adversarial attacks, which can prompt these systems to produce harmful responses. In the heart of these systems lies a safety classifier, a computational model trained to discern and mitigate potentially harmful, offensive, or unethical outputs. However, contemporary safety classifiers, despite their potential, often fail when exposed to inputs infused with adversarial noise. In response, our study introduces the Adversarial Prompt Shield (APS), a lightweight model that excels in detection accuracy and demon"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.00172","kind":"arxiv","version":1},"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/2311.00172/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":"2026-07-05T07:07:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"E1DWLD7Fnz85hfGBPpBw61RZtS3k6FAiLPqTbW+yfoaw9vjhYlkQmCm7UbfBtl29RYd5HHqsJtOqXFCvjM/rAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T07:39:12.938366Z"},"content_sha256":"7a950e5c66dcddda523550008e088f7d2dec8af30ffdebc469f014247ec49028","schema_version":"1.0","event_id":"sha256:7a950e5c66dcddda523550008e088f7d2dec8af30ffdebc469f014247ec49028"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/RTGGN4I6CAOIN5IQTZFLUOCXPZ/bundle.json","state_url":"https://pith.science/pith/RTGGN4I6CAOIN5IQTZFLUOCXPZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/RTGGN4I6CAOIN5IQTZFLUOCXPZ/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":"2026-07-07T07:39:12Z","links":{"resolver":"https://pith.science/pith/RTGGN4I6CAOIN5IQTZFLUOCXPZ","bundle":"https://pith.science/pith/RTGGN4I6CAOIN5IQTZFLUOCXPZ/bundle.json","state":"https://pith.science/pith/RTGGN4I6CAOIN5IQTZFLUOCXPZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/RTGGN4I6CAOIN5IQTZFLUOCXPZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:RTGGN4I6CAOIN5IQTZFLUOCXPZ","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":"5fd58d9c9ad37bc5fd17e3cc097c452f4eaa20ab0b2cf77b5b71d12e6549572c","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-31T22:22:10Z","title_canon_sha256":"577bffaf64f07e67ca3d3fefaf5005df2f94c3b38488d66dd4531b3c300c1575"},"schema_version":"1.0","source":{"id":"2311.00172","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2311.00172","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"arxiv_version","alias_value":"2311.00172v1","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.00172","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"pith_short_12","alias_value":"RTGGN4I6CAOI","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"pith_short_16","alias_value":"RTGGN4I6CAOIN5IQ","created_at":"2026-07-05T07:07:43Z"},{"alias_kind":"pith_short_8","alias_value":"RTGGN4I6","created_at":"2026-07-05T07:07:43Z"}],"graph_snapshots":[{"event_id":"sha256:7a950e5c66dcddda523550008e088f7d2dec8af30ffdebc469f014247ec49028","target":"graph","created_at":"2026-07-05T07:07:43Z","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/2311.00172/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models' safety remains a critical concern due to their vulnerability to adversarial attacks, which can prompt these systems to produce harmful responses. In the heart of these systems lies a safety classifier, a computational model trained to discern and mitigate potentially harmful, offensive, or unethical outputs. However, contemporary safety classifiers, despite their potential, often fail when exposed to inputs infused with adversarial noise. In response, our study introduces the Adversarial Prompt Shield (APS), a lightweight model that excels in detection accuracy and demon","authors_text":"Ali Derakhshan, Ian G. Harris, Jinhwa Kim","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-31T22:22:10Z","title":"Robust Safety Classifier for Large Language Models: Adversarial Prompt Shield"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.00172","kind":"arxiv","version":1},"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":"sha256:16813928c3febb39de91423b7cb550737d127fa1dc2223ef1be2df7a67dc8b68","target":"record","created_at":"2026-07-05T07:07:43Z","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":"5fd58d9c9ad37bc5fd17e3cc097c452f4eaa20ab0b2cf77b5b71d12e6549572c","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-31T22:22:10Z","title_canon_sha256":"577bffaf64f07e67ca3d3fefaf5005df2f94c3b38488d66dd4531b3c300c1575"},"schema_version":"1.0","source":{"id":"2311.00172","kind":"arxiv","version":1}},"canonical_sha256":"8ccc66f11e101c86f5109e4aba38577e75945bc02d9c8eb812a63f0e376d8c81","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8ccc66f11e101c86f5109e4aba38577e75945bc02d9c8eb812a63f0e376d8c81","first_computed_at":"2026-07-05T07:07:43.212682Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:07:43.212682Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2gcxq27ZQziUL9/MoLnRMg0z7BOfoDcAq6nVNJeaRecbike0rMKzJq8PaV6xo/0pFMGEpeCgNi3OgFUsC8M1Bw==","signature_status":"signed_v1","signed_at":"2026-07-05T07:07:43.213189Z","signed_message":"canonical_sha256_bytes"},"source_id":"2311.00172","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:16813928c3febb39de91423b7cb550737d127fa1dc2223ef1be2df7a67dc8b68","sha256:7a950e5c66dcddda523550008e088f7d2dec8af30ffdebc469f014247ec49028"],"state_sha256":"cd1c5d54d78a345faf6e61a2c304fe5d575f07d3a3d3785db2518321af34abf2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"JQxgU8SxHJp6s9XzPa1qt0Eo2ZMwciwmXG2aPIxprJPaQAjip7wpqo8cz2dNPtGQieZt8XYYztZf1P5ZLBpdCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T07:39:12.941699Z","bundle_sha256":"a5e6f0d3981927db2f8d4967d7cb47f08e9aad1327ba33f07ef3dc63e0ff27d9"}}