{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:N7RIEC7NHOAA2VHXDVE3HLBBN4","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":"ffe4680d04e3884d55b6d8db9592feaedf4266b6f30cad336f0bd31a1fc600c9","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-03-20T21:53:56Z","title_canon_sha256":"80b90ac908675826d323c5440b4274031e9cfb0da274def7eb25f54d7e1d3e53"},"schema_version":"1.0","source":{"id":"2403.14725","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.14725","created_at":"2026-07-05T08:35:44Z"},{"alias_kind":"arxiv_version","alias_value":"2403.14725v2","created_at":"2026-07-05T08:35:44Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.14725","created_at":"2026-07-05T08:35:44Z"},{"alias_kind":"pith_short_12","alias_value":"N7RIEC7NHOAA","created_at":"2026-07-05T08:35:44Z"},{"alias_kind":"pith_short_16","alias_value":"N7RIEC7NHOAA2VHX","created_at":"2026-07-05T08:35:44Z"},{"alias_kind":"pith_short_8","alias_value":"N7RIEC7N","created_at":"2026-07-05T08:35:44Z"}],"graph_snapshots":[{"event_id":"sha256:19980c9697a58b04c83304e59c4a81af229f88601a19e4764649d9a030e9bf67","target":"graph","created_at":"2026-07-05T08:35:44Z","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/2403.14725/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The rise of \"jailbreak\" attacks on language models has led to a flurry of defenses aimed at preventing undesirable responses. We critically examine the two stages of the defense pipeline: (i) defining what constitutes unsafe outputs, and (ii) enforcing the definition via methods such as input processing or fine-tuning. To test the efficacy of existing enforcement mechanisms, we consider a simple and well-specified definition of unsafe outputs--outputs that contain the word \"purple\". Surprisingly, existing fine-tuning and input defenses fail on this simple problem, casting doubt on whether enfo","authors_text":"Aditi Raghunathan, Suhas Kotha, Taeyoun Kim","cross_cats":["cs.CL","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-03-20T21:53:56Z","title":"Testing the Limits of Jailbreaking Defenses with the Purple Problem"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.14725","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:aa20686c995e1c7bc396542b8ecc98d7df386cc7c937aba6af650aeade5a3785","target":"record","created_at":"2026-07-05T08:35:44Z","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":"ffe4680d04e3884d55b6d8db9592feaedf4266b6f30cad336f0bd31a1fc600c9","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-03-20T21:53:56Z","title_canon_sha256":"80b90ac908675826d323c5440b4274031e9cfb0da274def7eb25f54d7e1d3e53"},"schema_version":"1.0","source":{"id":"2403.14725","kind":"arxiv","version":2}},"canonical_sha256":"6fe2820bed3b800d54f71d49b3ac216f39e6ac1e458c0615e098a9d84be673b3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6fe2820bed3b800d54f71d49b3ac216f39e6ac1e458c0615e098a9d84be673b3","first_computed_at":"2026-07-05T08:35:44.539586Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:35:44.539586Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"GnHGwlIhDO8ihdUBfQEU5PTqJUHfK4ua/qZ9DQGKq19Z8pjr2zuQ+Ad58ScnqY1ikG+zFu7Fi1HfLVKD/dcaAA==","signature_status":"signed_v1","signed_at":"2026-07-05T08:35:44.540115Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.14725","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:aa20686c995e1c7bc396542b8ecc98d7df386cc7c937aba6af650aeade5a3785","sha256:19980c9697a58b04c83304e59c4a81af229f88601a19e4764649d9a030e9bf67"],"state_sha256":"17bb7fa545c01463561feb5739b7496aadba9f8318b3ad5c29970b025668c035"}