{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2XI2WGRO55Z3YZRCIIFCHF7HHR","short_pith_number":"pith:2XI2WGRO","schema_version":"1.0","canonical_sha256":"d5d1ab1a2eef73bc6622420a2397e73c50db544ec5919f0dfcc69d1fcbc7d189","source":{"kind":"arxiv","id":"2504.10277","version":1},"attestation_state":"computed","paper":{"title":"RealHarm: A Collection of Real-World Language Model Application Failures","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CR"],"primary_cat":"cs.CY","authors_text":"Jiaen Liu, Luca Rossi, Matteo Dora, Pierre Le Jeune","submitted_at":"2025-04-14T14:44:41Z","abstract_excerpt":"Language model deployments in consumer-facing applications introduce numerous risks. While existing research on harms and hazards of such applications follows top-down approaches derived from regulatory frameworks and theoretical analyses, empirical evidence of real-world failure modes remains underexplored. In this work, we introduce RealHarm, a dataset of annotated problematic interactions with AI agents built from a systematic review of publicly reported incidents. Analyzing harms, causes, and hazards specifically from the deployer's perspective, we find that reputational damage constitutes"},"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":"2504.10277","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CY","submitted_at":"2025-04-14T14:44:41Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CR"],"title_canon_sha256":"0ed1d37c4290173f514bae31eede6cdab05055d38435616610e69e0582a13b0b","abstract_canon_sha256":"f2dd3b87e715c1f5bdaa3eddf081fa057fc0b263646690d6cfead04c393ae3d4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:07:06.014363Z","signature_b64":"uU6W4FJ96Q4y7kGXWHxr8IyE6mSjC6vfHUwF0SdkeR8k3BLdUisy9utnonVJSDOGrZyqdo976BVTB9Bzr3/ABw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d5d1ab1a2eef73bc6622420a2397e73c50db544ec5919f0dfcc69d1fcbc7d189","last_reissued_at":"2026-07-05T12:07:06.013768Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:07:06.013768Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RealHarm: A Collection of Real-World Language Model Application Failures","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CR"],"primary_cat":"cs.CY","authors_text":"Jiaen Liu, Luca Rossi, Matteo Dora, Pierre Le Jeune","submitted_at":"2025-04-14T14:44:41Z","abstract_excerpt":"Language model deployments in consumer-facing applications introduce numerous risks. While existing research on harms and hazards of such applications follows top-down approaches derived from regulatory frameworks and theoretical analyses, empirical evidence of real-world failure modes remains underexplored. In this work, we introduce RealHarm, a dataset of annotated problematic interactions with AI agents built from a systematic review of publicly reported incidents. Analyzing harms, causes, and hazards specifically from the deployer's perspective, we find that reputational damage constitutes"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.10277","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/2504.10277/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":"2504.10277","created_at":"2026-07-05T12:07:06.013835+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.10277v1","created_at":"2026-07-05T12:07:06.013835+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.10277","created_at":"2026-07-05T12:07:06.013835+00:00"},{"alias_kind":"pith_short_12","alias_value":"2XI2WGRO55Z3","created_at":"2026-07-05T12:07:06.013835+00:00"},{"alias_kind":"pith_short_16","alias_value":"2XI2WGRO55Z3YZRC","created_at":"2026-07-05T12:07:06.013835+00:00"},{"alias_kind":"pith_short_8","alias_value":"2XI2WGRO","created_at":"2026-07-05T12:07:06.013835+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.10442","citing_title":"StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10442","citing_title":"StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2604.28053","citing_title":"To Build or Not to Build? Factors that Lead to Non-Development or Abandonment of AI Systems","ref_index":80,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR","json":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR.json","graph_json":"https://pith.science/api/pith-number/2XI2WGRO55Z3YZRCIIFCHF7HHR/graph.json","events_json":"https://pith.science/api/pith-number/2XI2WGRO55Z3YZRCIIFCHF7HHR/events.json","paper":"https://pith.science/paper/2XI2WGRO"},"agent_actions":{"view_html":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR","download_json":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR.json","view_paper":"https://pith.science/paper/2XI2WGRO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.10277&json=true","fetch_graph":"https://pith.science/api/pith-number/2XI2WGRO55Z3YZRCIIFCHF7HHR/graph.json","fetch_events":"https://pith.science/api/pith-number/2XI2WGRO55Z3YZRCIIFCHF7HHR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR/action/storage_attestation","attest_author":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR/action/author_attestation","sign_citation":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR/action/citation_signature","submit_replication":"https://pith.science/pith/2XI2WGRO55Z3YZRCIIFCHF7HHR/action/replication_record"}},"created_at":"2026-07-05T12:07:06.013835+00:00","updated_at":"2026-07-05T12:07:06.013835+00:00"}