{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XKPTRRBV3SDVMRRUM4UXOT77AN","short_pith_number":"pith:XKPTRRBV","schema_version":"1.0","canonical_sha256":"ba9f38c435dc875646346729774fff035b2b6d4a417487577b975e20736b4a0d","source":{"kind":"arxiv","id":"2505.07846","version":1},"attestation_state":"computed","paper":{"title":"Winning at All Cost: A Small Environment for Eliciting Specification Gaming Behaviors in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.AI","authors_text":"Lars Malmqvist","submitted_at":"2025-05-07T07:59:56Z","abstract_excerpt":"This study reveals how frontier Large Language Models LLMs can \"game the system\" when faced with impossible situations, a critical security and alignment concern. Using a novel textual simulation approach, we presented three leading LLMs (o1, o3-mini, and r1) with a tic-tac-toe scenario designed to be unwinnable through legitimate play, then analyzed their tendency to exploit loopholes rather than accept defeat. Our results are alarming for security researchers: the newer, reasoning-focused o3-mini model showed nearly twice the propensity to exploit system vulnerabilities (37.1%) compared to t"},"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":"2505.07846","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-05-07T07:59:56Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"d27dd51878502cd13abaaee3eb36d187ac368f72b77b7a488770571206333b53","abstract_canon_sha256":"097a60900ed2bcd259b0590d6746c692701c6aa56c796db1483d9a8628c22fd3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:01:57.960662Z","signature_b64":"G9S4flHmMNS/LDifBvGaJXMqc+JZsw7SyClGKG/hzb2V8qcf2R2a+m4itBYm/+YhRu3fYANk2sxFBms1W9jwBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ba9f38c435dc875646346729774fff035b2b6d4a417487577b975e20736b4a0d","last_reissued_at":"2026-07-05T11:01:57.960206Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:01:57.960206Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Winning at All Cost: A Small Environment for Eliciting Specification Gaming Behaviors in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.AI","authors_text":"Lars Malmqvist","submitted_at":"2025-05-07T07:59:56Z","abstract_excerpt":"This study reveals how frontier Large Language Models LLMs can \"game the system\" when faced with impossible situations, a critical security and alignment concern. Using a novel textual simulation approach, we presented three leading LLMs (o1, o3-mini, and r1) with a tic-tac-toe scenario designed to be unwinnable through legitimate play, then analyzed their tendency to exploit loopholes rather than accept defeat. Our results are alarming for security researchers: the newer, reasoning-focused o3-mini model showed nearly twice the propensity to exploit system vulnerabilities (37.1%) compared to t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.07846","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/2505.07846/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":"2505.07846","created_at":"2026-07-05T11:01:57.960252+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.07846v1","created_at":"2026-07-05T11:01:57.960252+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.07846","created_at":"2026-07-05T11:01:57.960252+00:00"},{"alias_kind":"pith_short_12","alias_value":"XKPTRRBV3SDV","created_at":"2026-07-05T11:01:57.960252+00:00"},{"alias_kind":"pith_short_16","alias_value":"XKPTRRBV3SDVMRRU","created_at":"2026-07-05T11:01:57.960252+00:00"},{"alias_kind":"pith_short_8","alias_value":"XKPTRRBV","created_at":"2026-07-05T11:01:57.960252+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/XKPTRRBV3SDVMRRUM4UXOT77AN","json":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN.json","graph_json":"https://pith.science/api/pith-number/XKPTRRBV3SDVMRRUM4UXOT77AN/graph.json","events_json":"https://pith.science/api/pith-number/XKPTRRBV3SDVMRRUM4UXOT77AN/events.json","paper":"https://pith.science/paper/XKPTRRBV"},"agent_actions":{"view_html":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN","download_json":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN.json","view_paper":"https://pith.science/paper/XKPTRRBV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.07846&json=true","fetch_graph":"https://pith.science/api/pith-number/XKPTRRBV3SDVMRRUM4UXOT77AN/graph.json","fetch_events":"https://pith.science/api/pith-number/XKPTRRBV3SDVMRRUM4UXOT77AN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN/action/storage_attestation","attest_author":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN/action/author_attestation","sign_citation":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN/action/citation_signature","submit_replication":"https://pith.science/pith/XKPTRRBV3SDVMRRUM4UXOT77AN/action/replication_record"}},"created_at":"2026-07-05T11:01:57.960252+00:00","updated_at":"2026-07-05T11:01:57.960252+00:00"}