{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:HL3WGLCYM3257BRM3CZGI3PHKM","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":"16c48fbdbab7339f3b120c67ea97ff12a042d9b8c55c86a6f52e1ca2a3ee32f5","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2026-06-05T14:49:26Z","title_canon_sha256":"2685cd61bc6052a1d6b9fe6010de49bbed4301253e41c40ff2ffd66bed641cea"},"schema_version":"1.0","source":{"id":"2606.07335","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.07335","created_at":"2026-06-08T01:05:20Z"},{"alias_kind":"arxiv_version","alias_value":"2606.07335v1","created_at":"2026-06-08T01:05:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07335","created_at":"2026-06-08T01:05:20Z"},{"alias_kind":"pith_short_12","alias_value":"HL3WGLCYM325","created_at":"2026-06-08T01:05:20Z"},{"alias_kind":"pith_short_16","alias_value":"HL3WGLCYM3257BRM","created_at":"2026-06-08T01:05:20Z"},{"alias_kind":"pith_short_8","alias_value":"HL3WGLCY","created_at":"2026-06-08T01:05:20Z"}],"graph_snapshots":[{"event_id":"sha256:0723ec53a26819cd8453af1bdaeca03d9ea2f04d638eef75fb6f4d3bbc77f623","target":"graph","created_at":"2026-06-08T01:05:20Z","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/2606.07335/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Jailbreak prompts can bypass alignment guardrails in large language models (LLMs) and elicit unsafe outputs, making reliable deployment-time detection critical. Prior detection approaches largely rely on a fixed metric space, e.g., raw inputs, gradients, or hidden features, in which benign and jailbreak prompts are linearly separable. We show this assumption breaks under (i) pseudo-malicious prompts that are benign by intent but contain safety-related keywords, and (ii) adaptive attacks that explicitly optimize against the deployed detector. To overcome this limitation, we shift our focus from","authors_text":"Hangtao Zhang, Leo Yu Zhang, Minghui Li, Shengshan Hu, Sishun Liu, Wei Wan, Yanjun Zhang, Yi Liu, Yucheng Zhao, Zeyu Ye, Ziqi Zhou","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2026-06-05T14:49:26Z","title":"Defending Jailbreak Attacks on Large Language Models via Manifold Trajectory Kinetics"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07335","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:d8273e5b97ae71dbe9d835527848a1cd2fa6717026a9838f7bb692eb1cd4b2d4","target":"record","created_at":"2026-06-08T01:05:20Z","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":"16c48fbdbab7339f3b120c67ea97ff12a042d9b8c55c86a6f52e1ca2a3ee32f5","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2026-06-05T14:49:26Z","title_canon_sha256":"2685cd61bc6052a1d6b9fe6010de49bbed4301253e41c40ff2ffd66bed641cea"},"schema_version":"1.0","source":{"id":"2606.07335","kind":"arxiv","version":1}},"canonical_sha256":"3af7632c5866f5df862cd8b2646de75300ef93dd4118cef64a21f3044af40387","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3af7632c5866f5df862cd8b2646de75300ef93dd4118cef64a21f3044af40387","first_computed_at":"2026-06-08T01:05:20.396572Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-08T01:05:20.396572Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"s3R0P4mJ5iOEtOcWbJrnaEdw0C/X4c71C/KkHekvEBTwR6e5GHV3vOPykCDIRuGmFddtVEMQwa81vEdUj9EdCQ==","signature_status":"signed_v1","signed_at":"2026-06-08T01:05:20.397330Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.07335","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d8273e5b97ae71dbe9d835527848a1cd2fa6717026a9838f7bb692eb1cd4b2d4","sha256:0723ec53a26819cd8453af1bdaeca03d9ea2f04d638eef75fb6f4d3bbc77f623"],"state_sha256":"84833f541c7aa0458d8ce59decd3f728c1f63d5ba71d96183720a4386ef509fb"}