{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:6U5A3QHG3BWA6DJUOYFMYYAHCE","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":"0fdaf3152a837e042551840a625a08fdb24c0384c402ad11b19a323e0229be31","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-07-06T12:19:04Z","title_canon_sha256":"ab02d8bb814720a7b226e305c14fb0ab888920f9227e7c9a64c2c524ba9505d2"},"schema_version":"1.0","source":{"id":"2507.04365","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.04365","created_at":"2026-07-05T11:32:46Z"},{"alias_kind":"arxiv_version","alias_value":"2507.04365v1","created_at":"2026-07-05T11:32:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.04365","created_at":"2026-07-05T11:32:46Z"},{"alias_kind":"pith_short_12","alias_value":"6U5A3QHG3BWA","created_at":"2026-07-05T11:32:46Z"},{"alias_kind":"pith_short_16","alias_value":"6U5A3QHG3BWA6DJU","created_at":"2026-07-05T11:32:46Z"},{"alias_kind":"pith_short_8","alias_value":"6U5A3QHG","created_at":"2026-07-05T11:32:46Z"}],"graph_snapshots":[{"event_id":"sha256:76202848ffb3272d9ef566a1dcd1733c82f00a8dd50d1acb3c29517bca428137","target":"graph","created_at":"2026-07-05T11:32:46Z","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/2507.04365/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"As large language models (LLMs) become more integral to society and technology, ensuring their safety becomes essential. Jailbreak attacks exploit vulnerabilities to bypass safety guardrails, posing a significant threat. However, the mechanisms enabling these attacks are not well understood. In this paper, we reveal a universal phenomenon that occurs during jailbreak attacks: Attention Slipping. During this phenomenon, the model gradually reduces the attention it allocates to unsafe requests in a user query during the attack process, ultimately causing a jailbreak. We show Attention Slipping i","authors_text":"Pin-Yu Chen, Tsung-Yi Ho, Xiaomeng Hu","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-07-06T12:19:04Z","title":"Attention Slipping: A Mechanistic Understanding of Jailbreak Attacks and Defenses in LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.04365","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:fc4c461fcba89f206b690ef28a0abe5ada6bd2ce655e68c4fa83dd2588798d8e","target":"record","created_at":"2026-07-05T11:32:46Z","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":"0fdaf3152a837e042551840a625a08fdb24c0384c402ad11b19a323e0229be31","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-07-06T12:19:04Z","title_canon_sha256":"ab02d8bb814720a7b226e305c14fb0ab888920f9227e7c9a64c2c524ba9505d2"},"schema_version":"1.0","source":{"id":"2507.04365","kind":"arxiv","version":1}},"canonical_sha256":"f53a0dc0e6d86c0f0d34760acc6007111678d9a8ab10c07d3bb62638d1fa4611","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f53a0dc0e6d86c0f0d34760acc6007111678d9a8ab10c07d3bb62638d1fa4611","first_computed_at":"2026-07-05T11:32:46.547783Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:32:46.547783Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9zA5noWr1mztLk9sy4O5eymIgL3evzaYWt/lrf13oI1aV0LrW7ObNOWSGDv/yH0Z1aBWFblI+qgnqYbo2QEqAw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:32:46.548157Z","signed_message":"canonical_sha256_bytes"},"source_id":"2507.04365","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:fc4c461fcba89f206b690ef28a0abe5ada6bd2ce655e68c4fa83dd2588798d8e","sha256:76202848ffb3272d9ef566a1dcd1733c82f00a8dd50d1acb3c29517bca428137"],"state_sha256":"df7271d47aeb75d263c84de08bf8beabc4220ec0efc45a3e5a966350db0ae78a"}