{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LMPN6NSJ75J75FYKQ24TU2PC6K","short_pith_number":"pith:LMPN6NSJ","schema_version":"1.0","canonical_sha256":"5b1edf3649ff53fe970a86b93a69e2f296f3946266681b72826bf4a94b2a73e5","source":{"kind":"arxiv","id":"2505.19933","version":1},"attestation_state":"computed","paper":{"title":"Subtle Risks, Critical Failures: A Framework for Diagnosing Physical Safety of LLMs for Embodied Decision Making","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chanyoung Park, DongJu Jang, Jian Kim, Minseo Kim, Seungju Han, Sungwoong Kim, Yejin Son, Youngjae Yu","submitted_at":"2025-05-26T13:01:14Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly used for decision making in embodied agents, yet existing safety evaluations often rely on coarse success rates and domain-specific setups, making it difficult to diagnose why and where these models fail. This obscures our understanding of embodied safety and limits the selective deployment of LLMs in high-risk physical environments. We introduce SAFEL, the framework for systematically evaluating the physical safety of LLMs in embodied decision making. SAFEL assesses two key competencies: (1) rejecting unsafe commands via the Command Refusal Test, "},"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":"2505.19933","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-05-26T13:01:14Z","cross_cats_sorted":[],"title_canon_sha256":"5d9913faa2bfba76bd7f639177beba2277e521a0cb7cd0101011aad1a292a5ca","abstract_canon_sha256":"ba744f1422c1875704718bcc59220bafe1ee15a9226d7e1eb3b5f92e9e973096"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:50.081658Z","signature_b64":"eQnR/MGj/jSjvWlr1pudCfM6qUQCTypdwEpQN8mFkFbny+X73aKIfdabeSkck2MWQfR/5r69I22c0CHDnYmaAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5b1edf3649ff53fe970a86b93a69e2f296f3946266681b72826bf4a94b2a73e5","last_reissued_at":"2026-07-05T11:09:50.081188Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:50.081188Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Subtle Risks, Critical Failures: A Framework for Diagnosing Physical Safety of LLMs for Embodied Decision Making","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chanyoung Park, DongJu Jang, Jian Kim, Minseo Kim, Seungju Han, Sungwoong Kim, Yejin Son, Youngjae Yu","submitted_at":"2025-05-26T13:01:14Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly used for decision making in embodied agents, yet existing safety evaluations often rely on coarse success rates and domain-specific setups, making it difficult to diagnose why and where these models fail. This obscures our understanding of embodied safety and limits the selective deployment of LLMs in high-risk physical environments. We introduce SAFEL, the framework for systematically evaluating the physical safety of LLMs in embodied decision making. SAFEL assesses two key competencies: (1) rejecting unsafe commands via the Command Refusal Test, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19933","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/2505.19933/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":"2505.19933","created_at":"2026-07-05T11:09:50.081240+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.19933v1","created_at":"2026-07-05T11:09:50.081240+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19933","created_at":"2026-07-05T11:09:50.081240+00:00"},{"alias_kind":"pith_short_12","alias_value":"LMPN6NSJ75J7","created_at":"2026-07-05T11:09:50.081240+00:00"},{"alias_kind":"pith_short_16","alias_value":"LMPN6NSJ75J75FYK","created_at":"2026-07-05T11:09:50.081240+00:00"},{"alias_kind":"pith_short_8","alias_value":"LMPN6NSJ","created_at":"2026-07-05T11:09:50.081240+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/LMPN6NSJ75J75FYKQ24TU2PC6K","json":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K.json","graph_json":"https://pith.science/api/pith-number/LMPN6NSJ75J75FYKQ24TU2PC6K/graph.json","events_json":"https://pith.science/api/pith-number/LMPN6NSJ75J75FYKQ24TU2PC6K/events.json","paper":"https://pith.science/paper/LMPN6NSJ"},"agent_actions":{"view_html":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K","download_json":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K.json","view_paper":"https://pith.science/paper/LMPN6NSJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.19933&json=true","fetch_graph":"https://pith.science/api/pith-number/LMPN6NSJ75J75FYKQ24TU2PC6K/graph.json","fetch_events":"https://pith.science/api/pith-number/LMPN6NSJ75J75FYKQ24TU2PC6K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K/action/storage_attestation","attest_author":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K/action/author_attestation","sign_citation":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K/action/citation_signature","submit_replication":"https://pith.science/pith/LMPN6NSJ75J75FYKQ24TU2PC6K/action/replication_record"}},"created_at":"2026-07-05T11:09:50.081240+00:00","updated_at":"2026-07-05T11:09:50.081240+00:00"}