{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:GF5IUTBPOAECQY7ERMI3B5YITJ","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":"cd7733d7a2019524f72913f1bd62bf12c0cdf527ca81f479320cd33071ada2b5","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-03T05:00:12Z","title_canon_sha256":"b9d30e4fb898c0d101f217fa4f91a9828ca4fe962a2d14b1df862e25a2096fbf"},"schema_version":"1.0","source":{"id":"2506.02442","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.02442","created_at":"2026-07-05T11:15:47Z"},{"alias_kind":"arxiv_version","alias_value":"2506.02442v2","created_at":"2026-07-05T11:15:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.02442","created_at":"2026-07-05T11:15:47Z"},{"alias_kind":"pith_short_12","alias_value":"GF5IUTBPOAEC","created_at":"2026-07-05T11:15:47Z"},{"alias_kind":"pith_short_16","alias_value":"GF5IUTBPOAECQY7E","created_at":"2026-07-05T11:15:47Z"},{"alias_kind":"pith_short_8","alias_value":"GF5IUTBP","created_at":"2026-07-05T11:15:47Z"}],"graph_snapshots":[{"event_id":"sha256:b7c123d712239541fb073f387529481f6822cc9d3c6a3a33228fe508826a2ca6","target":"graph","created_at":"2026-07-05T11:15:47Z","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/2506.02442/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"This paper presents a systematic evaluation of Large Language Models' (LLMs) behavior on long-tail distributed (encrypted) texts and their safety implications. We introduce a two-dimensional framework for assessing LLM safety: (1) instruction refusal-the ability to reject harmful obfuscated instructions, and (2) generation safety-the suppression of generating harmful responses. Through comprehensive experiments, we demonstrate that models that possess capabilities to decrypt ciphers may be susceptible to mismatched-generalization attacks: their safety mechanisms fail on at least one safety dim","authors_text":"Mark Dras, Usman Naseem, Utsav Maskey","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-03T05:00:12Z","title":"Should LLM Safety Be More Than Refusing Harmful Instructions?"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.02442","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":"sha256:5329ec61031b324d700f1dfde9931086c4713a702f6eeaf27579827f20fc65e9","target":"record","created_at":"2026-07-05T11:15:47Z","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":"cd7733d7a2019524f72913f1bd62bf12c0cdf527ca81f479320cd33071ada2b5","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-03T05:00:12Z","title_canon_sha256":"b9d30e4fb898c0d101f217fa4f91a9828ca4fe962a2d14b1df862e25a2096fbf"},"schema_version":"1.0","source":{"id":"2506.02442","kind":"arxiv","version":2}},"canonical_sha256":"317a8a4c2f70082863e48b11b0f7089a4a789f6c51ae874b57af497686fd2f8c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"317a8a4c2f70082863e48b11b0f7089a4a789f6c51ae874b57af497686fd2f8c","first_computed_at":"2026-07-05T11:15:47.827417Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:15:47.827417Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"hQAVqxR8y4Q36ecmAJ+MyVjGvfcWAkUPVEKobRDWz6yQ4U1VQEMOogSM78jaPmgB6/ZjxW6Z5WYVMybH/48sAw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:15:47.827983Z","signed_message":"canonical_sha256_bytes"},"source_id":"2506.02442","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5329ec61031b324d700f1dfde9931086c4713a702f6eeaf27579827f20fc65e9","sha256:b7c123d712239541fb073f387529481f6822cc9d3c6a3a33228fe508826a2ca6"],"state_sha256":"28653f3e1bd76767aecccf1fa5f1e87b9368438e409b970235fd2d61b2237e69"}