{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7V5YWVTUHELDWSX2VUUQD3UWVL","short_pith_number":"pith:7V5YWVTU","schema_version":"1.0","canonical_sha256":"fd7b8b567439163b4afaad2901ee96aafb85a2123217aa3009fed9f4ce41317d","source":{"kind":"arxiv","id":"2605.19722","version":1},"attestation_state":"computed","paper":{"title":"Measuring Safety Alignment Effects in Autonomous Security Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Arthur Gervais, Isaac David","submitted_at":"2026-05-19T11:55:54Z","abstract_excerpt":"Do stock safety-aligned language models and their uncensored or abliterated derivatives behave differently when run as autonomous security agents? Single-turn refusal benchmarks cannot answer this question: security agents must inspect repositories, call tools, and produce vulnerability evidence inside authorized sandboxes.\n  We present a trace-based benchmark of 30 local vulnerability-analysis tasks with fixed tools, deterministic success predicates, redaction rules, and grounding checks, and compare four stock models against uncensored or abliterated derivatives: Gemma 4 31B, Gemma 4 26B A4B"},"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":"2605.19722","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-19T11:55:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"a6869ce9555745b95b87b377bb91ad3789a3aee1f38ac32b3318da8a1b8a1bb3","abstract_canon_sha256":"8cb34ac6d15eaa5ee24562f52f3cd3d88cf9f3aec919ff95d77ca05abf1866c8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:06:10.976447Z","signature_b64":"v64lADZBU8lrzMvl1WUKPnr4fYIy7UvP+0bUi9NvRIYnbzgFxDCBk6YR5CfHtMUypeYM7sqrLzdZse/gjQyfCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fd7b8b567439163b4afaad2901ee96aafb85a2123217aa3009fed9f4ce41317d","last_reissued_at":"2026-05-20T01:06:10.975824Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:06:10.975824Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Measuring Safety Alignment Effects in Autonomous Security Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Arthur Gervais, Isaac David","submitted_at":"2026-05-19T11:55:54Z","abstract_excerpt":"Do stock safety-aligned language models and their uncensored or abliterated derivatives behave differently when run as autonomous security agents? Single-turn refusal benchmarks cannot answer this question: security agents must inspect repositories, call tools, and produce vulnerability evidence inside authorized sandboxes.\n  We present a trace-based benchmark of 30 local vulnerability-analysis tasks with fixed tools, deterministic success predicates, redaction rules, and grounding checks, and compare four stock models against uncensored or abliterated derivatives: Gemma 4 31B, Gemma 4 26B A4B"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19722","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/2605.19722/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":"2605.19722","created_at":"2026-05-20T01:06:10.975921+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.19722v1","created_at":"2026-05-20T01:06:10.975921+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19722","created_at":"2026-05-20T01:06:10.975921+00:00"},{"alias_kind":"pith_short_12","alias_value":"7V5YWVTUHELD","created_at":"2026-05-20T01:06:10.975921+00:00"},{"alias_kind":"pith_short_16","alias_value":"7V5YWVTUHELDWSX2","created_at":"2026-05-20T01:06:10.975921+00:00"},{"alias_kind":"pith_short_8","alias_value":"7V5YWVTU","created_at":"2026-05-20T01:06:10.975921+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/7V5YWVTUHELDWSX2VUUQD3UWVL","json":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL.json","graph_json":"https://pith.science/api/pith-number/7V5YWVTUHELDWSX2VUUQD3UWVL/graph.json","events_json":"https://pith.science/api/pith-number/7V5YWVTUHELDWSX2VUUQD3UWVL/events.json","paper":"https://pith.science/paper/7V5YWVTU"},"agent_actions":{"view_html":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL","download_json":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL.json","view_paper":"https://pith.science/paper/7V5YWVTU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.19722&json=true","fetch_graph":"https://pith.science/api/pith-number/7V5YWVTUHELDWSX2VUUQD3UWVL/graph.json","fetch_events":"https://pith.science/api/pith-number/7V5YWVTUHELDWSX2VUUQD3UWVL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL/action/storage_attestation","attest_author":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL/action/author_attestation","sign_citation":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL/action/citation_signature","submit_replication":"https://pith.science/pith/7V5YWVTUHELDWSX2VUUQD3UWVL/action/replication_record"}},"created_at":"2026-05-20T01:06:10.975921+00:00","updated_at":"2026-05-20T01:06:10.975921+00:00"}