{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LTLRO5OA3DXRDTAPHYOLA4WMCA","short_pith_number":"pith:LTLRO5OA","schema_version":"1.0","canonical_sha256":"5cd71775c0d8ef11cc0f3e1cb072cc1005e15498310c1a2da4a044dd3d1b31e7","source":{"kind":"arxiv","id":"2506.00911","version":1},"attestation_state":"computed","paper":{"title":"Conformal Arbitrage: Risk-Controlled Balancing of Competing Objectives in Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Mohsen Bayati, William Overman","submitted_at":"2025-06-01T08:55:10Z","abstract_excerpt":"Modern language model deployments must often balance competing objectives, for example, helpfulness versus harmlessness, cost versus accuracy, and reward versus safety. We introduce Conformal Arbitrage, a post hoc framework that learns a data driven threshold to mediate between a Primary model optimized for a primary objective and a more conservative Guardian which could be another model or a human domain expert aligned with a guardrail objective. The threshold is calibrated with conformal risk control, yielding finite sample, distribution free guarantees that the long run frequency of undesir"},"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":"2506.00911","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-06-01T08:55:10Z","cross_cats_sorted":[],"title_canon_sha256":"bbe633a4611c8b4643ea2492d20a886937151a8f21c87e5bc0601aab4d5d051e","abstract_canon_sha256":"5cc5d0c3c596ccbc4345626c6bebb8e18e80f7256b1ef93c1fc2a268119b0376"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:13:42.680385Z","signature_b64":"GHoRRRnujx1moUqkhIKqYX1DoW1obsluKYPZpyvrvJs5qtmc7R09WaX09AkmomdgLytXTLSVJxh8SKSY5Dq2Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5cd71775c0d8ef11cc0f3e1cb072cc1005e15498310c1a2da4a044dd3d1b31e7","last_reissued_at":"2026-07-05T11:13:42.679896Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:13:42.679896Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Conformal Arbitrage: Risk-Controlled Balancing of Competing Objectives in Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Mohsen Bayati, William Overman","submitted_at":"2025-06-01T08:55:10Z","abstract_excerpt":"Modern language model deployments must often balance competing objectives, for example, helpfulness versus harmlessness, cost versus accuracy, and reward versus safety. We introduce Conformal Arbitrage, a post hoc framework that learns a data driven threshold to mediate between a Primary model optimized for a primary objective and a more conservative Guardian which could be another model or a human domain expert aligned with a guardrail objective. The threshold is calibrated with conformal risk control, yielding finite sample, distribution free guarantees that the long run frequency of undesir"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.00911","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/2506.00911/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":"2506.00911","created_at":"2026-07-05T11:13:42.679966+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.00911v1","created_at":"2026-07-05T11:13:42.679966+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.00911","created_at":"2026-07-05T11:13:42.679966+00:00"},{"alias_kind":"pith_short_12","alias_value":"LTLRO5OA3DXR","created_at":"2026-07-05T11:13:42.679966+00:00"},{"alias_kind":"pith_short_16","alias_value":"LTLRO5OA3DXRDTAP","created_at":"2026-07-05T11:13:42.679966+00:00"},{"alias_kind":"pith_short_8","alias_value":"LTLRO5OA","created_at":"2026-07-05T11:13:42.679966+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12587","citing_title":"Strategic Decision Support for AI Agents","ref_index":60,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA","json":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA.json","graph_json":"https://pith.science/api/pith-number/LTLRO5OA3DXRDTAPHYOLA4WMCA/graph.json","events_json":"https://pith.science/api/pith-number/LTLRO5OA3DXRDTAPHYOLA4WMCA/events.json","paper":"https://pith.science/paper/LTLRO5OA"},"agent_actions":{"view_html":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA","download_json":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA.json","view_paper":"https://pith.science/paper/LTLRO5OA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.00911&json=true","fetch_graph":"https://pith.science/api/pith-number/LTLRO5OA3DXRDTAPHYOLA4WMCA/graph.json","fetch_events":"https://pith.science/api/pith-number/LTLRO5OA3DXRDTAPHYOLA4WMCA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA/action/storage_attestation","attest_author":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA/action/author_attestation","sign_citation":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA/action/citation_signature","submit_replication":"https://pith.science/pith/LTLRO5OA3DXRDTAPHYOLA4WMCA/action/replication_record"}},"created_at":"2026-07-05T11:13:42.679966+00:00","updated_at":"2026-07-05T11:13:42.679966+00:00"}