{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IFMVGDYBKIGOLAY267AZXD272A","short_pith_number":"pith:IFMVGDYB","schema_version":"1.0","canonical_sha256":"4159530f01520ce5831af7c19b8f5fd01665cc8cd005abc1858b394803d029be","source":{"kind":"arxiv","id":"2507.18742","version":1},"attestation_state":"computed","paper":{"title":"Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"V\\'ictor Gallego","submitted_at":"2025-07-24T18:44:28Z","abstract_excerpt":"Language models (LMs) are susceptible to in-context reward hacking, where they exploit flaws in tainted or faulty written specifications or rubrics to achieve high scores without fulfilling the user's true intent. We introduce Specification Self-Correction (SSC), a novel, test-time framework that enables an LM to identify and correct flaws within its own guiding specification. SSC employs a multi-step inference process where the model first generates a response based on a potentially tainted specification, critiques its output, and then revises the specification itself to remove the exploitabl"},"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":"2507.18742","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-07-24T18:44:28Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"dba388cbd399e0cdb7e237d0df45ea23c2730fcc11d23cbaf525cf9dd53d8fd4","abstract_canon_sha256":"666154e42312cb28261b0614d512a9dcd15234a14cad9adc1a56d8fa3f1e754d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:42:57.132794Z","signature_b64":"eQZPSJzXZDs7inPtXLuV7pfbg1KUVc0NPPZsAMxjOn+SXDDAbrciU54OqJ++JhuXTaHTpLmMJ6LiwuZQseu8DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4159530f01520ce5831af7c19b8f5fd01665cc8cd005abc1858b394803d029be","last_reissued_at":"2026-07-05T11:42:57.132336Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:42:57.132336Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"V\\'ictor Gallego","submitted_at":"2025-07-24T18:44:28Z","abstract_excerpt":"Language models (LMs) are susceptible to in-context reward hacking, where they exploit flaws in tainted or faulty written specifications or rubrics to achieve high scores without fulfilling the user's true intent. We introduce Specification Self-Correction (SSC), a novel, test-time framework that enables an LM to identify and correct flaws within its own guiding specification. SSC employs a multi-step inference process where the model first generates a response based on a potentially tainted specification, critiques its output, and then revises the specification itself to remove the exploitabl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.18742","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/2507.18742/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":"2507.18742","created_at":"2026-07-05T11:42:57.132396+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.18742v1","created_at":"2026-07-05T11:42:57.132396+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.18742","created_at":"2026-07-05T11:42:57.132396+00:00"},{"alias_kind":"pith_short_12","alias_value":"IFMVGDYBKIGO","created_at":"2026-07-05T11:42:57.132396+00:00"},{"alias_kind":"pith_short_16","alias_value":"IFMVGDYBKIGOLAY2","created_at":"2026-07-05T11:42:57.132396+00:00"},{"alias_kind":"pith_short_8","alias_value":"IFMVGDYB","created_at":"2026-07-05T11:42:57.132396+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/IFMVGDYBKIGOLAY267AZXD272A","json":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A.json","graph_json":"https://pith.science/api/pith-number/IFMVGDYBKIGOLAY267AZXD272A/graph.json","events_json":"https://pith.science/api/pith-number/IFMVGDYBKIGOLAY267AZXD272A/events.json","paper":"https://pith.science/paper/IFMVGDYB"},"agent_actions":{"view_html":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A","download_json":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A.json","view_paper":"https://pith.science/paper/IFMVGDYB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.18742&json=true","fetch_graph":"https://pith.science/api/pith-number/IFMVGDYBKIGOLAY267AZXD272A/graph.json","fetch_events":"https://pith.science/api/pith-number/IFMVGDYBKIGOLAY267AZXD272A/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A/action/storage_attestation","attest_author":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A/action/author_attestation","sign_citation":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A/action/citation_signature","submit_replication":"https://pith.science/pith/IFMVGDYBKIGOLAY267AZXD272A/action/replication_record"}},"created_at":"2026-07-05T11:42:57.132396+00:00","updated_at":"2026-07-05T11:42:57.132396+00:00"}