{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZSVSSOCYOBE6TP3EPXE7F2RNRF","short_pith_number":"pith:ZSVSSOCY","schema_version":"1.0","canonical_sha256":"ccab2938587049e9bf647dc9f2ea2d89659801e32d51beffd863633101d02c26","source":{"kind":"arxiv","id":"2507.16126","version":1},"attestation_state":"computed","paper":{"title":"TaxCalcBench: Evaluating Frontier Models on the Tax Calculation Task","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Kara Molisee, Michael R. Bock, Sumit Shah, Zachary Ozer","submitted_at":"2025-07-22T00:37:59Z","abstract_excerpt":"Can AI file your taxes? Not yet. Calculating US personal income taxes is a task that requires building an understanding of vast amounts of English text and using that knowledge to carefully compute results. We propose TaxCalcBench, a benchmark for determining models' abilities to calculate personal income tax returns given all of the necessary information. Our experiment shows that state-of-the-art models succeed in calculating less than a third of federal income tax returns even on this simplified sample set. Our analysis concludes that models consistently misuse tax tables, make errors in ta"},"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":"2507.16126","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-07-22T00:37:59Z","cross_cats_sorted":[],"title_canon_sha256":"2b6fa7b399b9e780018461010433655ea0d686e8daf4c08ea246266b8eecc1e4","abstract_canon_sha256":"c2cbae66e8262f08e23291cb1cc56732099d1b31f184de978cb919ec2272b28c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:41:03.363277Z","signature_b64":"H/f/9ztESpmtN9eE+uc+P+jSSMJKCTbDiw1iNzeqZUYkq8uNzcEcXnRutZcwhmfpPHk8PZ02zIQ+veUwlZsrCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ccab2938587049e9bf647dc9f2ea2d89659801e32d51beffd863633101d02c26","last_reissued_at":"2026-07-05T11:41:03.362773Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:41:03.362773Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TaxCalcBench: Evaluating Frontier Models on the Tax Calculation Task","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Kara Molisee, Michael R. Bock, Sumit Shah, Zachary Ozer","submitted_at":"2025-07-22T00:37:59Z","abstract_excerpt":"Can AI file your taxes? Not yet. Calculating US personal income taxes is a task that requires building an understanding of vast amounts of English text and using that knowledge to carefully compute results. We propose TaxCalcBench, a benchmark for determining models' abilities to calculate personal income tax returns given all of the necessary information. Our experiment shows that state-of-the-art models succeed in calculating less than a third of federal income tax returns even on this simplified sample set. Our analysis concludes that models consistently misuse tax tables, make errors in ta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.16126","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/2507.16126/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":"2507.16126","created_at":"2026-07-05T11:41:03.362835+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.16126v1","created_at":"2026-07-05T11:41:03.362835+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.16126","created_at":"2026-07-05T11:41:03.362835+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZSVSSOCYOBE6","created_at":"2026-07-05T11:41:03.362835+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZSVSSOCYOBE6TP3E","created_at":"2026-07-05T11:41:03.362835+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZSVSSOCY","created_at":"2026-07-05T11:41:03.362835+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.08327","citing_title":"Interactive Critique-Revision Training for Reliable Structured LLM Generation","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF","json":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF.json","graph_json":"https://pith.science/api/pith-number/ZSVSSOCYOBE6TP3EPXE7F2RNRF/graph.json","events_json":"https://pith.science/api/pith-number/ZSVSSOCYOBE6TP3EPXE7F2RNRF/events.json","paper":"https://pith.science/paper/ZSVSSOCY"},"agent_actions":{"view_html":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF","download_json":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF.json","view_paper":"https://pith.science/paper/ZSVSSOCY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.16126&json=true","fetch_graph":"https://pith.science/api/pith-number/ZSVSSOCYOBE6TP3EPXE7F2RNRF/graph.json","fetch_events":"https://pith.science/api/pith-number/ZSVSSOCYOBE6TP3EPXE7F2RNRF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF/action/storage_attestation","attest_author":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF/action/author_attestation","sign_citation":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF/action/citation_signature","submit_replication":"https://pith.science/pith/ZSVSSOCYOBE6TP3EPXE7F2RNRF/action/replication_record"}},"created_at":"2026-07-05T11:41:03.362835+00:00","updated_at":"2026-07-05T11:41:03.362835+00:00"}