{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:C6DTQBMWHNIFR5Q276GZMXR7HT","short_pith_number":"pith:C6DTQBMW","schema_version":"1.0","canonical_sha256":"17873805963b5058f61aff8d965e3f3cd7c9f304e1f4ee97cbc3945937ca42e0","source":{"kind":"arxiv","id":"2509.02655","version":3},"attestation_state":"computed","paper":{"title":"BioBlue: Systematic runaway-optimiser-like LLM failure modes on biologically and economically aligned AI safety benchmarks for LLMs with simplified observation format","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CY","authors_text":"Roland Pihlakas, Sruthi Susan Kuriakose (for the Three Laws collaboration)","submitted_at":"2025-09-02T15:13:14Z","abstract_excerpt":"Many AI alignment discussions of \"runaway optimisation\" focus on RL agents: unbounded utility maximisers that over-optimise a proxy objective (e.g., \"paperclip maximiser\", specification gaming) at the expense of everything else. LLM-based systems are often assumed to be safer because they function as next-token predictors rather than persistent optimisers. We empirically test this assumption by placing LLMs in simple, long-horizon control-style environments that require maintaining state of or balancing objectives over time: single- and multi-objective homeostasis, balancing unbounded objectiv"},"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":"2509.02655","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CY","submitted_at":"2025-09-02T15:13:14Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"787fde48223698c3cfcd024f28a4d977ba642f9e3ea616e549cd3193422da5f9","abstract_canon_sha256":"6ce4faa9c7b32c47bbf02bd1579dda20f28278508cabb677640e3e8949998691"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-04T01:09:38.407986Z","signature_b64":"2s9tqstbeK+ac0xW8wwhNqezZnyp6QJnb6XiduN3VNyq0To9H4T/pHLqm0x7EOAFCgaQmSySY73hFYLhyC/5CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"17873805963b5058f61aff8d965e3f3cd7c9f304e1f4ee97cbc3945937ca42e0","last_reissued_at":"2026-06-04T01:09:38.406719Z","signature_status":"signed_v1","first_computed_at":"2026-06-04T01:09:38.406719Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BioBlue: Systematic runaway-optimiser-like LLM failure modes on biologically and economically aligned AI safety benchmarks for LLMs with simplified observation format","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CY","authors_text":"Roland Pihlakas, Sruthi Susan Kuriakose (for the Three Laws collaboration)","submitted_at":"2025-09-02T15:13:14Z","abstract_excerpt":"Many AI alignment discussions of \"runaway optimisation\" focus on RL agents: unbounded utility maximisers that over-optimise a proxy objective (e.g., \"paperclip maximiser\", specification gaming) at the expense of everything else. LLM-based systems are often assumed to be safer because they function as next-token predictors rather than persistent optimisers. We empirically test this assumption by placing LLMs in simple, long-horizon control-style environments that require maintaining state of or balancing objectives over time: single- and multi-objective homeostasis, balancing unbounded objectiv"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.02655","kind":"arxiv","version":3},"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/2509.02655/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":"2509.02655","created_at":"2026-06-04T01:09:38.406797+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.02655v3","created_at":"2026-06-04T01:09:38.406797+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.02655","created_at":"2026-06-04T01:09:38.406797+00:00"},{"alias_kind":"pith_short_12","alias_value":"C6DTQBMWHNIF","created_at":"2026-06-04T01:09:38.406797+00:00"},{"alias_kind":"pith_short_16","alias_value":"C6DTQBMWHNIFR5Q2","created_at":"2026-06-04T01:09:38.406797+00:00"},{"alias_kind":"pith_short_8","alias_value":"C6DTQBMW","created_at":"2026-06-04T01:09:38.406797+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/C6DTQBMWHNIFR5Q276GZMXR7HT","json":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT.json","graph_json":"https://pith.science/api/pith-number/C6DTQBMWHNIFR5Q276GZMXR7HT/graph.json","events_json":"https://pith.science/api/pith-number/C6DTQBMWHNIFR5Q276GZMXR7HT/events.json","paper":"https://pith.science/paper/C6DTQBMW"},"agent_actions":{"view_html":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT","download_json":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT.json","view_paper":"https://pith.science/paper/C6DTQBMW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.02655&json=true","fetch_graph":"https://pith.science/api/pith-number/C6DTQBMWHNIFR5Q276GZMXR7HT/graph.json","fetch_events":"https://pith.science/api/pith-number/C6DTQBMWHNIFR5Q276GZMXR7HT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT/action/storage_attestation","attest_author":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT/action/author_attestation","sign_citation":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT/action/citation_signature","submit_replication":"https://pith.science/pith/C6DTQBMWHNIFR5Q276GZMXR7HT/action/replication_record"}},"created_at":"2026-06-04T01:09:38.406797+00:00","updated_at":"2026-06-04T01:09:38.406797+00:00"}