{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:Q3RN7BXXCTBAJISIIHMQ3UOTP6","short_pith_number":"pith:Q3RN7BXX","canonical_record":{"source":{"id":"2608.09542","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-10T12:40:02Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"67051c7d8aba496c9ec9fd683fc232065b0e4f7ff367dcdd53e32da579e2b276","abstract_canon_sha256":"334cf6feacbc77fbb59f1a815d2f04132bd1c6547d5c89bdc2ae16f7b15af3b5"},"schema_version":"1.0"},"canonical_sha256":"86e2df86f714c204a24841d90dd1d37f935649f377ecfc9d9a2b287586dc5186","source":{"kind":"arxiv","id":"2608.09542","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.09542","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"arxiv_version","alias_value":"2608.09542v1","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.09542","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"pith_short_12","alias_value":"Q3RN7BXXCTBA","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"pith_short_16","alias_value":"Q3RN7BXXCTBAJISI","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"pith_short_8","alias_value":"Q3RN7BXX","created_at":"2026-08-11T02:24:27Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:Q3RN7BXXCTBAJISIIHMQ3UOTP6","target":"record","payload":{"canonical_record":{"source":{"id":"2608.09542","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-10T12:40:02Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"67051c7d8aba496c9ec9fd683fc232065b0e4f7ff367dcdd53e32da579e2b276","abstract_canon_sha256":"334cf6feacbc77fbb59f1a815d2f04132bd1c6547d5c89bdc2ae16f7b15af3b5"},"schema_version":"1.0"},"canonical_sha256":"86e2df86f714c204a24841d90dd1d37f935649f377ecfc9d9a2b287586dc5186","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T02:24:27.269024Z","signature_b64":"gCGfizBCu8WPslAeRh1qzD7o20ZlDfP+ayBCqON2UghZm/ruWDBVM6QgWweMH2ch+WibWfqaDW50R5DmcyipDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"86e2df86f714c204a24841d90dd1d37f935649f377ecfc9d9a2b287586dc5186","last_reissued_at":"2026-08-11T02:24:27.267478Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T02:24:27.267478Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2608.09542","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-08-11T02:24:27Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"IXup8/ZjmA/szG1r5NF54Y+MDQa3mAI0lpM/0BLtA41q5ZlyAXxagXY15TasFA7zorCkQ1oL3qZLFWoW+X78BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-16T21:11:03.253551Z"},"content_sha256":"003be856ee0e3f8ff71ad47cb8acdec157acf8ad1651440e7acf11cc698dadf4","schema_version":"1.0","event_id":"sha256:003be856ee0e3f8ff71ad47cb8acdec157acf8ad1651440e7acf11cc698dadf4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:Q3RN7BXXCTBAJISIIHMQ3UOTP6","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Dual-Adversarial Safety Alignment: Cultivating Intrinsic Threat Comprehension in LRMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CR"],"primary_cat":"cs.LG","authors_text":"Di Wang, Hongli Shen, Jian Li, Qinbo Zhang, Shaopeng Fu","submitted_at":"2026-08-10T12:40:02Z","abstract_excerpt":"Large reasoning models (LRMs) achieve remarkable success on complex tasks but remain vulnerable to harmful prompts that induce unsafe outputs. Recent methods align LRMs using direct refusals or safety rationales, yet often focus on prompt patterns rather than intrinsic attack mechanisms. As a result, these pattern-centric alignments struggle to generalize across diverse jailbreaks, compromising adversarial robustness and reasoning utility. We propose AdvSafe, a dual-adversarial framework that enables LRMs to internalize unsafety knowledge by explicitly deconstructing adversarial mechanisms. Th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.09542","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2608.09542/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-08-11T02:24:27Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"O7zeTya05FMQJtl73EuaXMsIxeuUJN3w8y8+XRrdIHJoP0Kdf73TLQ31cDJDeDRvx5O8r7qw9lBEDsHcAUqPCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-16T21:11:03.254143Z"},"content_sha256":"5195f8dfcaba08c1ed4dc44f5d9fcc838fee2c927735caed859c86646ed57c5d","schema_version":"1.0","event_id":"sha256:5195f8dfcaba08c1ed4dc44f5d9fcc838fee2c927735caed859c86646ed57c5d"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/Q3RN7BXXCTBAJISIIHMQ3UOTP6/bundle.json","state_url":"https://pith.science/pith/Q3RN7BXXCTBAJISIIHMQ3UOTP6/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/Q3RN7BXXCTBAJISIIHMQ3UOTP6/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-16T21:11:03Z","links":{"resolver":"https://pith.science/pith/Q3RN7BXXCTBAJISIIHMQ3UOTP6","bundle":"https://pith.science/pith/Q3RN7BXXCTBAJISIIHMQ3UOTP6/bundle.json","state":"https://pith.science/pith/Q3RN7BXXCTBAJISIIHMQ3UOTP6/state.json","well_known_bundle":"https://pith.science/.well-known/pith/Q3RN7BXXCTBAJISIIHMQ3UOTP6/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:Q3RN7BXXCTBAJISIIHMQ3UOTP6","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":"334cf6feacbc77fbb59f1a815d2f04132bd1c6547d5c89bdc2ae16f7b15af3b5","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-10T12:40:02Z","title_canon_sha256":"67051c7d8aba496c9ec9fd683fc232065b0e4f7ff367dcdd53e32da579e2b276"},"schema_version":"1.0","source":{"id":"2608.09542","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.09542","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"arxiv_version","alias_value":"2608.09542v1","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.09542","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"pith_short_12","alias_value":"Q3RN7BXXCTBA","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"pith_short_16","alias_value":"Q3RN7BXXCTBAJISI","created_at":"2026-08-11T02:24:27Z"},{"alias_kind":"pith_short_8","alias_value":"Q3RN7BXX","created_at":"2026-08-11T02:24:27Z"}],"graph_snapshots":[{"event_id":"sha256:5195f8dfcaba08c1ed4dc44f5d9fcc838fee2c927735caed859c86646ed57c5d","target":"graph","created_at":"2026-08-11T02:24:27Z","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/2608.09542/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large reasoning models (LRMs) achieve remarkable success on complex tasks but remain vulnerable to harmful prompts that induce unsafe outputs. Recent methods align LRMs using direct refusals or safety rationales, yet often focus on prompt patterns rather than intrinsic attack mechanisms. As a result, these pattern-centric alignments struggle to generalize across diverse jailbreaks, compromising adversarial robustness and reasoning utility. We propose AdvSafe, a dual-adversarial framework that enables LRMs to internalize unsafety knowledge by explicitly deconstructing adversarial mechanisms. Th","authors_text":"Di Wang, Hongli Shen, Jian Li, Qinbo Zhang, Shaopeng Fu","cross_cats":["cs.AI","cs.CR"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-10T12:40:02Z","title":"Dual-Adversarial Safety Alignment: Cultivating Intrinsic Threat Comprehension in LRMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.09542","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":"sha256:003be856ee0e3f8ff71ad47cb8acdec157acf8ad1651440e7acf11cc698dadf4","target":"record","created_at":"2026-08-11T02:24:27Z","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":"334cf6feacbc77fbb59f1a815d2f04132bd1c6547d5c89bdc2ae16f7b15af3b5","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-10T12:40:02Z","title_canon_sha256":"67051c7d8aba496c9ec9fd683fc232065b0e4f7ff367dcdd53e32da579e2b276"},"schema_version":"1.0","source":{"id":"2608.09542","kind":"arxiv","version":1}},"canonical_sha256":"86e2df86f714c204a24841d90dd1d37f935649f377ecfc9d9a2b287586dc5186","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"86e2df86f714c204a24841d90dd1d37f935649f377ecfc9d9a2b287586dc5186","first_computed_at":"2026-08-11T02:24:27.267478Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-08-11T02:24:27.267478Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"gCGfizBCu8WPslAeRh1qzD7o20ZlDfP+ayBCqON2UghZm/ruWDBVM6QgWweMH2ch+WibWfqaDW50R5DmcyipDA==","signature_status":"signed_v1","signed_at":"2026-08-11T02:24:27.269024Z","signed_message":"canonical_sha256_bytes"},"source_id":"2608.09542","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:003be856ee0e3f8ff71ad47cb8acdec157acf8ad1651440e7acf11cc698dadf4","sha256:5195f8dfcaba08c1ed4dc44f5d9fcc838fee2c927735caed859c86646ed57c5d"],"state_sha256":"3af93564a0039bcf49b4dc29bc5d3a968ca6722cd0580f4ec6d90c9b78315180"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Wb+sD9ile6PP4HYE9HDc5r2MckgV98ZnKENIR6EGlmcAJDRII9zNsbntH5sXZCHvHZiPe+Dnq/RRE6j4Nf4eBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-16T21:11:03.258517Z","bundle_sha256":"9894157c6e21d49dcf8b58bea61fd07e25d7e022cf6224074a5cbba62ca5344f"}}