{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:2CK553QXFT5TWNFILSBIXNHTHN","short_pith_number":"pith:2CK553QX","schema_version":"1.0","canonical_sha256":"d095deee172cfb3b34a85c828bb4f33b452e8b7ec9d75476a2bee2ab40630df1","source":{"kind":"arxiv","id":"2606.04717","version":1},"attestation_state":"computed","paper":{"title":"Selection-Aware Diagnostics for Chain-of-Thought Answer Hijacking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CY"],"primary_cat":"cs.CR","authors_text":"Jianwei Tai","submitted_at":"2026-06-03T10:49:43Z","abstract_excerpt":"We study a controlled numeric proxy for chain-of-thought (CoT) answer hijacking, motivated by attacks in which benign-looking reasoning steers a harmful final answer. CoT wrappers on GSM8K and MATH-500 flip final answers away from gold labels. Rather than treating activation patching as clean-trace restoration, we ask where hijacked trajectories are fragile and whether recovery depends on a same-problem clean source. Across Qwen2.5-7B and Llama3-8B on GSM8K few-shot, puzzle, and sycophant hijacks, three few-shot/puzzle cells pass confirmatory $K{=}1$ localization after Bonferroni correction. A"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.04717","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-06-03T10:49:43Z","cross_cats_sorted":["cs.CY"],"title_canon_sha256":"54d5c101dcab2e31b78d5f71d7a34229fd6c1c7eb6f1f2c85c7bbd6215528ce4","abstract_canon_sha256":"dad808c554fcaa1d964f2e0e044098526036af5d22ea079571f9ff37b53a5c1f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-04T01:09:26.444500Z","signature_b64":"KnvUDSUlMnVWDto72ZuqENP3BDqVuV6715hnsL6IQxxxD1o2FETtlDKH/zo1AiLYrJue3RU9LApCX7JNzwhfBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d095deee172cfb3b34a85c828bb4f33b452e8b7ec9d75476a2bee2ab40630df1","last_reissued_at":"2026-06-04T01:09:26.443738Z","signature_status":"signed_v1","first_computed_at":"2026-06-04T01:09:26.443738Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Selection-Aware Diagnostics for Chain-of-Thought Answer Hijacking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CY"],"primary_cat":"cs.CR","authors_text":"Jianwei Tai","submitted_at":"2026-06-03T10:49:43Z","abstract_excerpt":"We study a controlled numeric proxy for chain-of-thought (CoT) answer hijacking, motivated by attacks in which benign-looking reasoning steers a harmful final answer. CoT wrappers on GSM8K and MATH-500 flip final answers away from gold labels. Rather than treating activation patching as clean-trace restoration, we ask where hijacked trajectories are fragile and whether recovery depends on a same-problem clean source. Across Qwen2.5-7B and Llama3-8B on GSM8K few-shot, puzzle, and sycophant hijacks, three few-shot/puzzle cells pass confirmatory $K{=}1$ localization after Bonferroni correction. A"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.04717","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/2606.04717/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.04717","created_at":"2026-06-04T01:09:26.443873+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.04717v1","created_at":"2026-06-04T01:09:26.443873+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.04717","created_at":"2026-06-04T01:09:26.443873+00:00"},{"alias_kind":"pith_short_12","alias_value":"2CK553QXFT5T","created_at":"2026-06-04T01:09:26.443873+00:00"},{"alias_kind":"pith_short_16","alias_value":"2CK553QXFT5TWNFI","created_at":"2026-06-04T01:09:26.443873+00:00"},{"alias_kind":"pith_short_8","alias_value":"2CK553QX","created_at":"2026-06-04T01:09:26.443873+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN","json":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN.json","graph_json":"https://pith.science/api/pith-number/2CK553QXFT5TWNFILSBIXNHTHN/graph.json","events_json":"https://pith.science/api/pith-number/2CK553QXFT5TWNFILSBIXNHTHN/events.json","paper":"https://pith.science/paper/2CK553QX"},"agent_actions":{"view_html":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN","download_json":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN.json","view_paper":"https://pith.science/paper/2CK553QX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.04717&json=true","fetch_graph":"https://pith.science/api/pith-number/2CK553QXFT5TWNFILSBIXNHTHN/graph.json","fetch_events":"https://pith.science/api/pith-number/2CK553QXFT5TWNFILSBIXNHTHN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN/action/storage_attestation","attest_author":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN/action/author_attestation","sign_citation":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN/action/citation_signature","submit_replication":"https://pith.science/pith/2CK553QXFT5TWNFILSBIXNHTHN/action/replication_record"}},"created_at":"2026-06-04T01:09:26.443873+00:00","updated_at":"2026-06-04T01:09:26.443873+00:00"}