{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:YDOTUXDKCOJ2VSOKV7ENL4O5QY","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":"b8ccf03db52eb7d8aa86d7fd7a0115626fd11618248eed8d1bfcc2f8dfb49500","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-06T18:55:18Z","title_canon_sha256":"7a0857d817bb2bb942d775ca81034a98fce6901ada1df14bf425bfbc9de18366"},"schema_version":"1.0","source":{"id":"2311.03348","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2311.03348","created_at":"2026-07-05T07:16:02Z"},{"alias_kind":"arxiv_version","alias_value":"2311.03348v2","created_at":"2026-07-05T07:16:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.03348","created_at":"2026-07-05T07:16:02Z"},{"alias_kind":"pith_short_12","alias_value":"YDOTUXDKCOJ2","created_at":"2026-07-05T07:16:02Z"},{"alias_kind":"pith_short_16","alias_value":"YDOTUXDKCOJ2VSOK","created_at":"2026-07-05T07:16:02Z"},{"alias_kind":"pith_short_8","alias_value":"YDOTUXDK","created_at":"2026-07-05T07:16:02Z"}],"graph_snapshots":[{"event_id":"sha256:6e212cdde537f07812966de29103845526710ecc4ddb2ddecefe0f66b0de8771","target":"graph","created_at":"2026-07-05T07:16:02Z","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/2311.03348/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Despite efforts to align large language models to produce harmless responses, they are still vulnerable to jailbreak prompts that elicit unrestricted behaviour. In this work, we investigate persona modulation as a black-box jailbreaking method to steer a target model to take on personalities that are willing to comply with harmful instructions. Rather than manually crafting prompts for each persona, we automate the generation of jailbreaks using a language model assistant. We demonstrate a range of harmful completions made possible by persona modulation, including detailed instructions for syn","authors_text":"Arush Tagade, Javier Rando, Quentin Feuillade--Montixi, Rusheb Shah, Soroush Pour, Stephen Casper","cross_cats":["cs.AI","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-06T18:55:18Z","title":"Scalable and Transferable Black-Box Jailbreaks for Language Models via Persona Modulation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.03348","kind":"arxiv","version":2},"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":"sha256:37fb04186212e68a33bcb1fdc2f7f8872294f65054e90572c547fef27cce038a","target":"record","created_at":"2026-07-05T07:16:02Z","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":"b8ccf03db52eb7d8aa86d7fd7a0115626fd11618248eed8d1bfcc2f8dfb49500","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-06T18:55:18Z","title_canon_sha256":"7a0857d817bb2bb942d775ca81034a98fce6901ada1df14bf425bfbc9de18366"},"schema_version":"1.0","source":{"id":"2311.03348","kind":"arxiv","version":2}},"canonical_sha256":"c0dd3a5c6a1393aac9caafc8d5f1dd8627b36a6125e5d58357767e8562fdbf91","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c0dd3a5c6a1393aac9caafc8d5f1dd8627b36a6125e5d58357767e8562fdbf91","first_computed_at":"2026-07-05T07:16:02.521459Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:16:02.521459Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"/NNZfPRQ/l/f98b3IGAWAD2frNbN2DyPoExb/hq9PFz/7ZHyxScOnwknXhbfRrhCjXnDwdcez6YZnxex+sATBA==","signature_status":"signed_v1","signed_at":"2026-07-05T07:16:02.521924Z","signed_message":"canonical_sha256_bytes"},"source_id":"2311.03348","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:37fb04186212e68a33bcb1fdc2f7f8872294f65054e90572c547fef27cce038a","sha256:6e212cdde537f07812966de29103845526710ecc4ddb2ddecefe0f66b0de8771"],"state_sha256":"a7744f152f5e7ec04a52c41e2a21f684d8c17690e6df0f9224bd35d20ce09ddf"}