{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:L5IHLUCBEDTXWFQ3HINUOLIVPW","short_pith_number":"pith:L5IHLUCB","schema_version":"1.0","canonical_sha256":"5f5075d04120e77b161b3a1b472d157db02cb79e29388dab8364ca18f8cf2cfa","source":{"kind":"arxiv","id":"2606.01066","version":1},"attestation_state":"computed","paper":{"title":"Before the Model Learns the Bug:Fuzzing RLVR Verifiers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Jaideep Ray","submitted_at":"2026-05-31T07:18:07Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) replaces human preference labels with executable reward functions such as math answer checkers, JSON tool-call validators, and code unit-test harnesses. That makes the reward partly a software artifact: if the verifier is wrong, optimization can learn the bug. We study this failure mode with a lightweight verifier-fuzzing framework that generates adversarial completions, compares buggy and stricter reference verifiers, logs paired decisions, and reports false-positive, false-negative, disagreement, exploit, and uncertainty metrics."},"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":"2606.01066","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-31T07:18:07Z","cross_cats_sorted":[],"title_canon_sha256":"93b3a9d241a0aa5e14635560e7e9508ab3f657e724682b65e946937da02a3b39","abstract_canon_sha256":"599f87103d250cf9e05ed757cd2b5d5ab979ef231cb7682a5ef4a3dbfe7f00f7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:04:20.323698Z","signature_b64":"zk+K/+8mnPUKy8cuP1AxIQYbstNIm58+/akopWLIyODa3MdkPcaFW457jnaglTG2oLhlYzM1P8Igcju25go5AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5f5075d04120e77b161b3a1b472d157db02cb79e29388dab8364ca18f8cf2cfa","last_reissued_at":"2026-06-02T01:04:20.323143Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:04:20.323143Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Before the Model Learns the Bug:Fuzzing RLVR Verifiers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Jaideep Ray","submitted_at":"2026-05-31T07:18:07Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) replaces human preference labels with executable reward functions such as math answer checkers, JSON tool-call validators, and code unit-test harnesses. That makes the reward partly a software artifact: if the verifier is wrong, optimization can learn the bug. We study this failure mode with a lightweight verifier-fuzzing framework that generates adversarial completions, compares buggy and stricter reference verifiers, logs paired decisions, and reports false-positive, false-negative, disagreement, exploit, and uncertainty metrics."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01066","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/2606.01066/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":"2606.01066","created_at":"2026-06-02T01:04:20.323252+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01066v1","created_at":"2026-06-02T01:04:20.323252+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01066","created_at":"2026-06-02T01:04:20.323252+00:00"},{"alias_kind":"pith_short_12","alias_value":"L5IHLUCBEDTX","created_at":"2026-06-02T01:04:20.323252+00:00"},{"alias_kind":"pith_short_16","alias_value":"L5IHLUCBEDTXWFQ3","created_at":"2026-06-02T01:04:20.323252+00:00"},{"alias_kind":"pith_short_8","alias_value":"L5IHLUCB","created_at":"2026-06-02T01:04:20.323252+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/L5IHLUCBEDTXWFQ3HINUOLIVPW","json":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW.json","graph_json":"https://pith.science/api/pith-number/L5IHLUCBEDTXWFQ3HINUOLIVPW/graph.json","events_json":"https://pith.science/api/pith-number/L5IHLUCBEDTXWFQ3HINUOLIVPW/events.json","paper":"https://pith.science/paper/L5IHLUCB"},"agent_actions":{"view_html":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW","download_json":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW.json","view_paper":"https://pith.science/paper/L5IHLUCB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01066&json=true","fetch_graph":"https://pith.science/api/pith-number/L5IHLUCBEDTXWFQ3HINUOLIVPW/graph.json","fetch_events":"https://pith.science/api/pith-number/L5IHLUCBEDTXWFQ3HINUOLIVPW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW/action/storage_attestation","attest_author":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW/action/author_attestation","sign_citation":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW/action/citation_signature","submit_replication":"https://pith.science/pith/L5IHLUCBEDTXWFQ3HINUOLIVPW/action/replication_record"}},"created_at":"2026-06-02T01:04:20.323252+00:00","updated_at":"2026-06-02T01:04:20.323252+00:00"}