{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:SKSWLDU5V2B4CA6PVZTHMXQR6V","short_pith_number":"pith:SKSWLDU5","schema_version":"1.0","canonical_sha256":"92a5658e9dae83c103cfae66765e11f56cf7d18337fa62240552294a32fdbfe2","source":{"kind":"arxiv","id":"2502.04512","version":4},"attestation_state":"computed","paper":{"title":"Safety Must Precede the Deployment of Open-Ended AI","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Open-ended AI systems pose unique safety challenges that existing methods cannot address and must be tackled before deployment.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Ivaxi Sheth, Jan Wehner, Mario Fritz, Ruta Binkyte, Sahar Abdelnabi","submitted_at":"2025-02-06T21:32:07Z","abstract_excerpt":"AI advancements have been significantly driven by a combination of foundation models and curiosity-driven learning aimed at increasing capability and adaptability. Within this landscape, open-endedness, where AI agents autonomously and indefinitely generate novel behaviors, representations, or solutions, has gained increasing interest. This has become relevant in the context of self-evolving agents and long-horizon discovery. This position paper argues that the defining properties of open-ended AI systems introduce a distinct and underexplored class of safety challenges, including loss of pred"},"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":"2502.04512","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-02-06T21:32:07Z","cross_cats_sorted":[],"title_canon_sha256":"311e477417bf8f3fb540ed83450f43ee50703aaf8c6e60956d324e6a4a3989c7","abstract_canon_sha256":"7323d5defb202971ef47035e71e47f3b2819f21b79565daaeffe2d8df42ab986"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:46.286425Z","signature_b64":"E3n/VbunIIiDE1hgBtNG3LN9/SKHbf2gdxHPzNnIgmmQQ36NmqDebyux50kpIcc1h8B6OpIOQm5gNZZYbrt4AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"92a5658e9dae83c103cfae66765e11f56cf7d18337fa62240552294a32fdbfe2","last_reissued_at":"2026-06-02T02:04:46.285910Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:46.285910Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Safety Must Precede the Deployment of Open-Ended AI","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Open-ended AI systems pose unique safety challenges that existing methods cannot address and must be tackled before deployment.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Ivaxi Sheth, Jan Wehner, Mario Fritz, Ruta Binkyte, Sahar Abdelnabi","submitted_at":"2025-02-06T21:32:07Z","abstract_excerpt":"AI advancements have been significantly driven by a combination of foundation models and curiosity-driven learning aimed at increasing capability and adaptability. Within this landscape, open-endedness, where AI agents autonomously and indefinitely generate novel behaviors, representations, or solutions, has gained increasing interest. This has become relevant in the context of self-evolving agents and long-horizon discovery. This position paper argues that the defining properties of open-ended AI systems introduce a distinct and underexplored class of safety challenges, including loss of pred"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"the defining properties of open-ended AI systems introduce a distinct and underexplored class of safety challenges, including loss of predictability, emergent misalignment, and difficulties in maintaining effective control as systems evolve beyond their initial design assumptions, that must be addressed preemptively.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"These challenges differ qualitatively from those associated with task-bounded or static models and are unlikely to be addressed by existing safety frameworks alone.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Open-ended AI systems introduce distinct safety challenges including loss of predictability, emergent misalignment, and control difficulties that must be addressed preemptively before large-scale deployment.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Open-ended AI systems pose unique safety challenges that existing methods cannot address and must be tackled before deployment.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0e677e6b00ee6ad1be965e49e0cd5b56e3ee461298c0ab658e71c964cb00a62e"},"source":{"id":"2502.04512","kind":"arxiv","version":4},"verdict":{"id":"9d339520-15b3-4c56-9bae-b5766023ea6b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-23T03:24:01.912960Z","strongest_claim":"the defining properties of open-ended AI systems introduce a distinct and underexplored class of safety challenges, including loss of predictability, emergent misalignment, and difficulties in maintaining effective control as systems evolve beyond their initial design assumptions, that must be addressed preemptively.","one_line_summary":"Open-ended AI systems introduce distinct safety challenges including loss of predictability, emergent misalignment, and control difficulties that must be addressed preemptively before large-scale deployment.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"These challenges differ qualitatively from those associated with task-bounded or static models and are unlikely to be addressed by existing safety frameworks alone.","pith_extraction_headline":"Open-ended AI systems pose unique safety challenges that existing methods cannot address and must be tackled before deployment."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2502.04512/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":"2502.04512","created_at":"2026-06-02T02:04:46.285973+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.04512v4","created_at":"2026-06-02T02:04:46.285973+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.04512","created_at":"2026-06-02T02:04:46.285973+00:00"},{"alias_kind":"pith_short_12","alias_value":"SKSWLDU5V2B4","created_at":"2026-06-02T02:04:46.285973+00:00"},{"alias_kind":"pith_short_16","alias_value":"SKSWLDU5V2B4CA6P","created_at":"2026-06-02T02:04:46.285973+00:00"},{"alias_kind":"pith_short_8","alias_value":"SKSWLDU5","created_at":"2026-06-02T02:04:46.285973+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/SKSWLDU5V2B4CA6PVZTHMXQR6V","json":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V.json","graph_json":"https://pith.science/api/pith-number/SKSWLDU5V2B4CA6PVZTHMXQR6V/graph.json","events_json":"https://pith.science/api/pith-number/SKSWLDU5V2B4CA6PVZTHMXQR6V/events.json","paper":"https://pith.science/paper/SKSWLDU5"},"agent_actions":{"view_html":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V","download_json":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V.json","view_paper":"https://pith.science/paper/SKSWLDU5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.04512&json=true","fetch_graph":"https://pith.science/api/pith-number/SKSWLDU5V2B4CA6PVZTHMXQR6V/graph.json","fetch_events":"https://pith.science/api/pith-number/SKSWLDU5V2B4CA6PVZTHMXQR6V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V/action/storage_attestation","attest_author":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V/action/author_attestation","sign_citation":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V/action/citation_signature","submit_replication":"https://pith.science/pith/SKSWLDU5V2B4CA6PVZTHMXQR6V/action/replication_record"}},"created_at":"2026-06-02T02:04:46.285973+00:00","updated_at":"2026-06-02T02:04:46.285973+00:00"}