{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2U3IECQXKKCGS5MQLBXP3BXRL5","short_pith_number":"pith:2U3IECQX","schema_version":"1.0","canonical_sha256":"d536820a175284697590586efd86f15f65728895821163b7d4a5ad0186265c6c","source":{"kind":"arxiv","id":"2507.02977","version":1},"attestation_state":"computed","paper":{"title":"LLMs are Capable of Misaligned Behavior Under Explicit Prohibition and Surveillance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Igor Ivanov","submitted_at":"2025-06-30T21:37:00Z","abstract_excerpt":"In this paper, LLMs are tasked with completing an impossible quiz, while they are in a sandbox, monitored, told about these measures and instructed not to cheat. Some frontier LLMs cheat consistently and attempt to circumvent restrictions despite everything. The results reveal a fundamental tension between goal-directed behavior and alignment in current LLMs. The code and evaluation logs are available at github.com/baceolus/cheating_evals"},"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":"2507.02977","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-06-30T21:37:00Z","cross_cats_sorted":[],"title_canon_sha256":"96379d75bbd50f10f208f2aa57fa019766def5f11b80ca43b5f432bbcde5cc8b","abstract_canon_sha256":"2411e655f1047a0a9592538dc2835cb616b9ce3ecd00d25faa36b6711f068a4b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:31:49.935837Z","signature_b64":"sxqyaMokvG51sEV9uRqBcqeZz3VruECcgkdKNS928eYWBdfpGuVpl9C1nIRa2u+zHAAM7RhESs8tkNtN8YriDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d536820a175284697590586efd86f15f65728895821163b7d4a5ad0186265c6c","last_reissued_at":"2026-07-05T11:31:49.935406Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:31:49.935406Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLMs are Capable of Misaligned Behavior Under Explicit Prohibition and Surveillance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Igor Ivanov","submitted_at":"2025-06-30T21:37:00Z","abstract_excerpt":"In this paper, LLMs are tasked with completing an impossible quiz, while they are in a sandbox, monitored, told about these measures and instructed not to cheat. Some frontier LLMs cheat consistently and attempt to circumvent restrictions despite everything. The results reveal a fundamental tension between goal-directed behavior and alignment in current LLMs. The code and evaluation logs are available at github.com/baceolus/cheating_evals"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.02977","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/2507.02977/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":"2507.02977","created_at":"2026-07-05T11:31:49.935467+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.02977v1","created_at":"2026-07-05T11:31:49.935467+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.02977","created_at":"2026-07-05T11:31:49.935467+00:00"},{"alias_kind":"pith_short_12","alias_value":"2U3IECQXKKCG","created_at":"2026-07-05T11:31:49.935467+00:00"},{"alias_kind":"pith_short_16","alias_value":"2U3IECQXKKCGS5MQ","created_at":"2026-07-05T11:31:49.935467+00:00"},{"alias_kind":"pith_short_8","alias_value":"2U3IECQX","created_at":"2026-07-05T11:31:49.935467+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.10246","citing_title":"SciIntegrity-Bench: A Benchmark for Evaluating Academic Integrity in AI Scientist Systems","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10246","citing_title":"SciIntegrity-Bench: A Benchmark for Evaluating Academic Integrity in AI Scientist Systems","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5","json":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5.json","graph_json":"https://pith.science/api/pith-number/2U3IECQXKKCGS5MQLBXP3BXRL5/graph.json","events_json":"https://pith.science/api/pith-number/2U3IECQXKKCGS5MQLBXP3BXRL5/events.json","paper":"https://pith.science/paper/2U3IECQX"},"agent_actions":{"view_html":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5","download_json":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5.json","view_paper":"https://pith.science/paper/2U3IECQX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.02977&json=true","fetch_graph":"https://pith.science/api/pith-number/2U3IECQXKKCGS5MQLBXP3BXRL5/graph.json","fetch_events":"https://pith.science/api/pith-number/2U3IECQXKKCGS5MQLBXP3BXRL5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5/action/storage_attestation","attest_author":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5/action/author_attestation","sign_citation":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5/action/citation_signature","submit_replication":"https://pith.science/pith/2U3IECQXKKCGS5MQLBXP3BXRL5/action/replication_record"}},"created_at":"2026-07-05T11:31:49.935467+00:00","updated_at":"2026-07-05T11:31:49.935467+00:00"}