{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:AEOUYY32WHR2UHE6CTT2GNNVFS","short_pith_number":"pith:AEOUYY32","canonical_record":{"source":{"id":"2308.13768","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-26T05:20:58Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"22b1d8f030ad9a53715fe20008dc0e2407044b152bf828c23c42f3edd7919457","abstract_canon_sha256":"c583453f2c17077fd6181e31cdb779ea842dfc57d83d3fff03178ede32e6c5ee"},"schema_version":"1.0"},"canonical_sha256":"011d4c637ab1e3aa1c9e14e7a335b52ca5dd04218d7f38bbfe8e63a51f626003","source":{"kind":"arxiv","id":"2308.13768","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2308.13768","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"arxiv_version","alias_value":"2308.13768v1","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.13768","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"pith_short_12","alias_value":"AEOUYY32WHR2","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"pith_short_16","alias_value":"AEOUYY32WHR2UHE6","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"pith_short_8","alias_value":"AEOUYY32","created_at":"2026-07-05T06:45:04Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:AEOUYY32WHR2UHE6CTT2GNNVFS","target":"record","payload":{"canonical_record":{"source":{"id":"2308.13768","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-26T05:20:58Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"22b1d8f030ad9a53715fe20008dc0e2407044b152bf828c23c42f3edd7919457","abstract_canon_sha256":"c583453f2c17077fd6181e31cdb779ea842dfc57d83d3fff03178ede32e6c5ee"},"schema_version":"1.0"},"canonical_sha256":"011d4c637ab1e3aa1c9e14e7a335b52ca5dd04218d7f38bbfe8e63a51f626003","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:45:04.676292Z","signature_b64":"bINRFXPbaOu6WPsbgTKHJPWVWOyinyrJH9wgBhRqHmIuSMhbpvnLC8ZOeb0Vl9VKIb59akmOisvQybYoE0EEAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"011d4c637ab1e3aa1c9e14e7a335b52ca5dd04218d7f38bbfe8e63a51f626003","last_reissued_at":"2026-07-05T06:45:04.675848Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:45:04.675848Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2308.13768","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":"2026-07-05T06:45:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5TL4NXVGBkiDWW2JMZrPoV3aL0GHJMKgNvBzHQLhB1uyIAjDCpwtgGI9E7xciwk+m5QU0Lic7FEJQOEwriH2CA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T04:39:51.654124Z"},"content_sha256":"375fd77143aab314ffb5f0affd9565cfe729e47cd31acfba08fff32150923a93","schema_version":"1.0","event_id":"sha256:375fd77143aab314ffb5f0affd9565cfe729e47cd31acfba08fff32150923a93"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:AEOUYY32WHR2UHE6CTT2GNNVFS","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Adversarial Fine-Tuning of Language Models: An Iterative Optimisation Approach for the Generation and Detection of Problematic Content","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Charles O'Neill, Ioana Ciuca, Jack Miller, Thang Bui, Yuan-Sen Ting","submitted_at":"2023-08-26T05:20:58Z","abstract_excerpt":"In this paper, we tackle the emerging challenge of unintended harmful content generation in Large Language Models (LLMs) with a novel dual-stage optimisation technique using adversarial fine-tuning. Our two-pronged approach employs an adversarial model, fine-tuned to generate potentially harmful prompts, and a judge model, iteratively optimised to discern these prompts. In this adversarial cycle, the two models seek to outperform each other in the prompting phase, generating a dataset of rich examples which are then used for fine-tuning. This iterative application of prompting and fine-tuning "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.13768","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/2308.13768/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":"2026-07-05T06:45:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RbzaiGk35P64JFNGo7nXfJUmfiq0/pPXCPpZHhc3a6J1N0jMIHqC10J7g0fGX3zZYDmUwG9+V3BB5ogoLtrWAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T04:39:51.654625Z"},"content_sha256":"e8ec4e1da2cf08a1863b09c4c0106118e5ffbd781f2661814c446da82e0f97e5","schema_version":"1.0","event_id":"sha256:e8ec4e1da2cf08a1863b09c4c0106118e5ffbd781f2661814c446da82e0f97e5"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/AEOUYY32WHR2UHE6CTT2GNNVFS/bundle.json","state_url":"https://pith.science/pith/AEOUYY32WHR2UHE6CTT2GNNVFS/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/AEOUYY32WHR2UHE6CTT2GNNVFS/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":"2026-08-08T04:39:51Z","links":{"resolver":"https://pith.science/pith/AEOUYY32WHR2UHE6CTT2GNNVFS","bundle":"https://pith.science/pith/AEOUYY32WHR2UHE6CTT2GNNVFS/bundle.json","state":"https://pith.science/pith/AEOUYY32WHR2UHE6CTT2GNNVFS/state.json","well_known_bundle":"https://pith.science/.well-known/pith/AEOUYY32WHR2UHE6CTT2GNNVFS/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:AEOUYY32WHR2UHE6CTT2GNNVFS","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":"c583453f2c17077fd6181e31cdb779ea842dfc57d83d3fff03178ede32e6c5ee","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-26T05:20:58Z","title_canon_sha256":"22b1d8f030ad9a53715fe20008dc0e2407044b152bf828c23c42f3edd7919457"},"schema_version":"1.0","source":{"id":"2308.13768","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2308.13768","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"arxiv_version","alias_value":"2308.13768v1","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.13768","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"pith_short_12","alias_value":"AEOUYY32WHR2","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"pith_short_16","alias_value":"AEOUYY32WHR2UHE6","created_at":"2026-07-05T06:45:04Z"},{"alias_kind":"pith_short_8","alias_value":"AEOUYY32","created_at":"2026-07-05T06:45:04Z"}],"graph_snapshots":[{"event_id":"sha256:e8ec4e1da2cf08a1863b09c4c0106118e5ffbd781f2661814c446da82e0f97e5","target":"graph","created_at":"2026-07-05T06:45:04Z","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/2308.13768/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In this paper, we tackle the emerging challenge of unintended harmful content generation in Large Language Models (LLMs) with a novel dual-stage optimisation technique using adversarial fine-tuning. Our two-pronged approach employs an adversarial model, fine-tuned to generate potentially harmful prompts, and a judge model, iteratively optimised to discern these prompts. In this adversarial cycle, the two models seek to outperform each other in the prompting phase, generating a dataset of rich examples which are then used for fine-tuning. This iterative application of prompting and fine-tuning ","authors_text":"Charles O'Neill, Ioana Ciuca, Jack Miller, Thang Bui, Yuan-Sen Ting","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-26T05:20:58Z","title":"Adversarial Fine-Tuning of Language Models: An Iterative Optimisation Approach for the Generation and Detection of Problematic Content"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.13768","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":"sha256:375fd77143aab314ffb5f0affd9565cfe729e47cd31acfba08fff32150923a93","target":"record","created_at":"2026-07-05T06:45:04Z","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":"c583453f2c17077fd6181e31cdb779ea842dfc57d83d3fff03178ede32e6c5ee","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-26T05:20:58Z","title_canon_sha256":"22b1d8f030ad9a53715fe20008dc0e2407044b152bf828c23c42f3edd7919457"},"schema_version":"1.0","source":{"id":"2308.13768","kind":"arxiv","version":1}},"canonical_sha256":"011d4c637ab1e3aa1c9e14e7a335b52ca5dd04218d7f38bbfe8e63a51f626003","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"011d4c637ab1e3aa1c9e14e7a335b52ca5dd04218d7f38bbfe8e63a51f626003","first_computed_at":"2026-07-05T06:45:04.675848Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T06:45:04.675848Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"bINRFXPbaOu6WPsbgTKHJPWVWOyinyrJH9wgBhRqHmIuSMhbpvnLC8ZOeb0Vl9VKIb59akmOisvQybYoE0EEAw==","signature_status":"signed_v1","signed_at":"2026-07-05T06:45:04.676292Z","signed_message":"canonical_sha256_bytes"},"source_id":"2308.13768","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:375fd77143aab314ffb5f0affd9565cfe729e47cd31acfba08fff32150923a93","sha256:e8ec4e1da2cf08a1863b09c4c0106118e5ffbd781f2661814c446da82e0f97e5"],"state_sha256":"c1c6fda334a766245678fb350f9a89c5ab90d5aff172e52bad03ace980690f5e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"reCqYuLR9V9WnjKIzxGhuSlaZ8xJ+FOXocOi6HuuVDSgL1RB6zZSziRZZPnuv88peUIi+HI+geXvM/he6JQCBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T04:39:51.659405Z","bundle_sha256":"961a9023dad5ad1c58761316cd15736b27ec5539ea369d10df7bcb62182f15f9"}}