{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:H6A5RDXK4RK7IOAHBRCR5WAJNQ","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":"474bea5fa84658696e36f765576746199e00dfe91b28daa7f085f55aeab2f6ae","cross_cats_sorted":["cs.AI","cs.CL","cs.CY"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-02-01T22:26:30Z","title_canon_sha256":"b62ae633230ccf16db7d3da10e63ac81bc317caad3ddbb5b5e576a94c827cca8"},"schema_version":"1.0","source":{"id":"2502.00580","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.00580","created_at":"2026-07-05T10:08:35Z"},{"alias_kind":"arxiv_version","alias_value":"2502.00580v1","created_at":"2026-07-05T10:08:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.00580","created_at":"2026-07-05T10:08:35Z"},{"alias_kind":"pith_short_12","alias_value":"H6A5RDXK4RK7","created_at":"2026-07-05T10:08:35Z"},{"alias_kind":"pith_short_16","alias_value":"H6A5RDXK4RK7IOAH","created_at":"2026-07-05T10:08:35Z"},{"alias_kind":"pith_short_8","alias_value":"H6A5RDXK","created_at":"2026-07-05T10:08:35Z"}],"graph_snapshots":[{"event_id":"sha256:0b714adecaaca12e5f4168605d4c69da903b51a0812054c81535f378b3072bea","target":"graph","created_at":"2026-07-05T10:08:35Z","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/2502.00580/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent work showed Best-of-N (BoN) jailbreaking using repeated use of random augmentations (such as capitalization, punctuation, etc) is effective against all major large language models (LLMs). We have found that $100\\%$ of the BoN paper's successful jailbreaks (confidence interval $[99.65\\%, 100.00\\%]$) and $99.8\\%$ of successful jailbreaks in our replication (confidence interval $[99.28\\%, 99.98\\%]$) were blocked with our Defense Against The Dark Prompts (DATDP) method. The DATDP algorithm works by repeatedly utilizing an evaluation LLM to evaluate a prompt for dangerous or manipulative beh","authors_text":"Connor Stevens, Matija Franklin, Rebecca Gorman, Stuart Armstrong","cross_cats":["cs.AI","cs.CL","cs.CY"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-02-01T22:26:30Z","title":"Defense Against the Dark Prompts: Mitigating Best-of-N Jailbreaking with Prompt Evaluation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.00580","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:bc5db979ab79ca23bfa4a5bda677b0fdc110be7e12855f1d62f6c51b5529dacd","target":"record","created_at":"2026-07-05T10:08:35Z","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":"474bea5fa84658696e36f765576746199e00dfe91b28daa7f085f55aeab2f6ae","cross_cats_sorted":["cs.AI","cs.CL","cs.CY"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-02-01T22:26:30Z","title_canon_sha256":"b62ae633230ccf16db7d3da10e63ac81bc317caad3ddbb5b5e576a94c827cca8"},"schema_version":"1.0","source":{"id":"2502.00580","kind":"arxiv","version":1}},"canonical_sha256":"3f81d88eeae455f438070c451ed8096c37c054073e32acba08290bf89c11f212","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3f81d88eeae455f438070c451ed8096c37c054073e32acba08290bf89c11f212","first_computed_at":"2026-07-05T10:08:35.668724Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:08:35.668724Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Do4WHq8/VIFuOFPm3pxO/ZvBcingjRiJx50W3hvGrOTIX3Ly+FR/RSp3tnl1s82Q+Uiag7dUslxVTQwUg3U3DA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:08:35.669142Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.00580","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bc5db979ab79ca23bfa4a5bda677b0fdc110be7e12855f1d62f6c51b5529dacd","sha256:0b714adecaaca12e5f4168605d4c69da903b51a0812054c81535f378b3072bea"],"state_sha256":"8be9c39efaa97473c6c227797a968d7b42793757b7c9bb893fe68c98b3a2564f"}