{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:EDS4IEIJLTPD6OULJZTW5MCFWC","short_pith_number":"pith:EDS4IEIJ","schema_version":"1.0","canonical_sha256":"20e5c411095cde3f3a8b4e676eb045b09a1e11b4b53a68b256b60c7982016046","source":{"kind":"arxiv","id":"2605.17268","version":1},"attestation_state":"computed","paper":{"title":"Is VLA Reasoning Faithful? Probing Safety of Chain-of-Causation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV","cs.RO"],"primary_cat":"cs.AI","authors_text":"Nicanor Mayumu, Patrick Mukala, Xiaoheng Deng","submitted_at":"2026-05-17T05:29:48Z","abstract_excerpt":"We present the first systematic study of faithfulness in Vision-Language-Action (VLA) driving models, analyzing 300 Alpamayo-R1-10B inferences across 100 diverse PhysicalAI-AV scenarios. Our main finding is that output natural-language rationales with trajectories may be significantly unfaithful: (i) overall reasoning fidelity is only 42.5%, with Chain-of-Causation matching scene reality less than half the time; (ii) 94 missed pedestrians in one-third of pedestrian-relevant scenes; (iii) 97.7% trajectory fragility under mild visual perturbations; and (iv) only 48.3% mean reasoning-action consi"},"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":"2605.17268","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-17T05:29:48Z","cross_cats_sorted":["cs.CV","cs.RO"],"title_canon_sha256":"38e574f28dff6277896e67e42a1f39d7bf459a82b1ce24550bcf99fb03ff3e73","abstract_canon_sha256":"416940bfd8dfbe91a608a613ed86657de3462aeafc4a607ada53106853f5b928"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:03:48.907621Z","signature_b64":"R7yujIwcbib7UErqqVjm7XuPBQTgAC9JFKYkN2jtogakg8VO0HBtojoD5vY+uysVA0nFkkRimaplPtFjXZa3Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"20e5c411095cde3f3a8b4e676eb045b09a1e11b4b53a68b256b60c7982016046","last_reissued_at":"2026-05-20T00:03:48.906786Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:03:48.906786Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Is VLA Reasoning Faithful? Probing Safety of Chain-of-Causation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV","cs.RO"],"primary_cat":"cs.AI","authors_text":"Nicanor Mayumu, Patrick Mukala, Xiaoheng Deng","submitted_at":"2026-05-17T05:29:48Z","abstract_excerpt":"We present the first systematic study of faithfulness in Vision-Language-Action (VLA) driving models, analyzing 300 Alpamayo-R1-10B inferences across 100 diverse PhysicalAI-AV scenarios. Our main finding is that output natural-language rationales with trajectories may be significantly unfaithful: (i) overall reasoning fidelity is only 42.5%, with Chain-of-Causation matching scene reality less than half the time; (ii) 94 missed pedestrians in one-third of pedestrian-relevant scenes; (iii) 97.7% trajectory fragility under mild visual perturbations; and (iv) only 48.3% mean reasoning-action consi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.17268","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/2605.17268/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-19T22:01:57.840071Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.779687Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"23240c6957d2c654ee2a48718f0d4e4e9a1f25678b7130e02e422635a40eef49"},"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":"2605.17268","created_at":"2026-05-20T00:03:48.906926+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.17268v1","created_at":"2026-05-20T00:03:48.906926+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.17268","created_at":"2026-05-20T00:03:48.906926+00:00"},{"alias_kind":"pith_short_12","alias_value":"EDS4IEIJLTPD","created_at":"2026-05-20T00:03:48.906926+00:00"},{"alias_kind":"pith_short_16","alias_value":"EDS4IEIJLTPD6OUL","created_at":"2026-05-20T00:03:48.906926+00:00"},{"alias_kind":"pith_short_8","alias_value":"EDS4IEIJ","created_at":"2026-05-20T00:03:48.906926+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/EDS4IEIJLTPD6OULJZTW5MCFWC","json":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC.json","graph_json":"https://pith.science/api/pith-number/EDS4IEIJLTPD6OULJZTW5MCFWC/graph.json","events_json":"https://pith.science/api/pith-number/EDS4IEIJLTPD6OULJZTW5MCFWC/events.json","paper":"https://pith.science/paper/EDS4IEIJ"},"agent_actions":{"view_html":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC","download_json":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC.json","view_paper":"https://pith.science/paper/EDS4IEIJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.17268&json=true","fetch_graph":"https://pith.science/api/pith-number/EDS4IEIJLTPD6OULJZTW5MCFWC/graph.json","fetch_events":"https://pith.science/api/pith-number/EDS4IEIJLTPD6OULJZTW5MCFWC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC/action/storage_attestation","attest_author":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC/action/author_attestation","sign_citation":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC/action/citation_signature","submit_replication":"https://pith.science/pith/EDS4IEIJLTPD6OULJZTW5MCFWC/action/replication_record"}},"created_at":"2026-05-20T00:03:48.906926+00:00","updated_at":"2026-05-20T00:03:48.906926+00:00"}