{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:GMFJBMXXD32TYLTC6CQZTDH3AA","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":"d3c6ccfeb079a90f21d8eca1b6a852a83d177555f873e0ff32a2939a458a606b","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-10T00:35:23Z","title_canon_sha256":"9793c4fc1c6ef8cded44a19d5e978a2e6999f0ac5e4c6bd494a60f4d255772b8"},"schema_version":"1.0","source":{"id":"2406.05946","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.05946","created_at":"2026-07-05T08:29:35Z"},{"alias_kind":"arxiv_version","alias_value":"2406.05946v1","created_at":"2026-07-05T08:29:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.05946","created_at":"2026-07-05T08:29:35Z"},{"alias_kind":"pith_short_12","alias_value":"GMFJBMXXD32T","created_at":"2026-07-05T08:29:35Z"},{"alias_kind":"pith_short_16","alias_value":"GMFJBMXXD32TYLTC","created_at":"2026-07-05T08:29:35Z"},{"alias_kind":"pith_short_8","alias_value":"GMFJBMXX","created_at":"2026-07-05T08:29:35Z"}],"graph_snapshots":[{"event_id":"sha256:9dbb45200af793b49c65b8473dbf4a1e2274e06759a249c5066f28388cb9b229","target":"graph","created_at":"2026-07-05T08:29:35Z","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/2406.05946/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The safety alignment of current Large Language Models (LLMs) is vulnerable. Relatively simple attacks, or even benign fine-tuning, can jailbreak aligned models. We argue that many of these vulnerabilities are related to a shared underlying issue: safety alignment can take shortcuts, wherein the alignment adapts a model's generative distribution primarily over only its very first few output tokens. We refer to this issue as shallow safety alignment. In this paper, we present case studies to explain why shallow safety alignment can exist and provide evidence that current aligned LLMs are subject","authors_text":"Ahmad Beirami, Ashwinee Panda, Kaifeng Lyu, Peter Henderson, Prateek Mittal, Subhrajit Roy, Xiangyu Qi, Xiao Ma","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-10T00:35:23Z","title":"Safety Alignment Should Be Made More Than Just a Few Tokens Deep"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.05946","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":"sha256:9805f91870378d04421641cef880a164f6a8130d2687d7d2f1b9af81b1ef0f92","target":"record","created_at":"2026-07-05T08:29:35Z","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":"d3c6ccfeb079a90f21d8eca1b6a852a83d177555f873e0ff32a2939a458a606b","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-10T00:35:23Z","title_canon_sha256":"9793c4fc1c6ef8cded44a19d5e978a2e6999f0ac5e4c6bd494a60f4d255772b8"},"schema_version":"1.0","source":{"id":"2406.05946","kind":"arxiv","version":1}},"canonical_sha256":"330a90b2f71ef53c2e62f0a1998cfb00167774894c7ccb2dae23de76bd54fbd2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"330a90b2f71ef53c2e62f0a1998cfb00167774894c7ccb2dae23de76bd54fbd2","first_computed_at":"2026-07-05T08:29:35.633937Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:29:35.633937Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"iIBPi+5OaHDK324l1jdrxjdFTfzr7xWbk3LC5leCHDyaVQPJRpsQXedU6Ba8ypsASrnajCt7TE9UaVaHaDmwAg==","signature_status":"signed_v1","signed_at":"2026-07-05T08:29:35.634382Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.05946","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:9805f91870378d04421641cef880a164f6a8130d2687d7d2f1b9af81b1ef0f92","sha256:9dbb45200af793b49c65b8473dbf4a1e2274e06759a249c5066f28388cb9b229"],"state_sha256":"8c4dc467fddbd6c352f28eb080d9e7af6733b39ee2d2966a3b5bda5bff0d62e2"}