{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:XCU4YPGOOR3SVI3WUH5KTU63LL","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"2bf6e8b15b7d24b329c9fde9c61f0033e9a84f94e4b293723ed84bb72932d814","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-20T10:33:11Z","title_canon_sha256":"bd5135c0a3f26fa9c6cfe4cf850bd019a5809642d5f4c42f3f70055ebf521dd5"},"schema_version":"1.0","source":{"id":"2605.20994","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.20994","created_at":"2026-05-21T01:05:31Z"},{"alias_kind":"arxiv_version","alias_value":"2605.20994v1","created_at":"2026-05-21T01:05:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.20994","created_at":"2026-05-21T01:05:31Z"},{"alias_kind":"pith_short_12","alias_value":"XCU4YPGOOR3S","created_at":"2026-05-21T01:05:31Z"},{"alias_kind":"pith_short_16","alias_value":"XCU4YPGOOR3SVI3W","created_at":"2026-05-21T01:05:31Z"},{"alias_kind":"pith_short_8","alias_value":"XCU4YPGO","created_at":"2026-05-21T01:05:31Z"}],"graph_snapshots":[{"event_id":"sha256:6cfa3abc776091bb21ae4b3141983e635ebbada40c89de96efc02aa1cff2285a","target":"graph","created_at":"2026-05-21T01:05:31Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2605.20994/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Preference-based post-training aligns LLMs with human intent, yet safety behavior often remains brittle. A model may refuse a harmful request in a standard prompt but comply when the same intent is wrapped in adversarial wording. We suggest that robust safety requires context-invariant alignment, where behavior depends on the underlying intent rather than surface form. Enforcing invariance is difficult in alignment because not all training signals are equally trustworthy; for some prompt variants we can obtain verifiable feedback (e.g., multiple-choice), while for open-ended variants we typica","authors_text":"Xingjun Ma, Xin Wang, Yang Yao, Yan Teng, Yifeng Gao, Yingchun Wang, Yixu Wang","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-20T10:33:11Z","title":"Towards Context-Invariant Safety Alignment for Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.20994","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:80f969be053e8bf862e5f3ce9feefebac59c8fd85135853794eb07db87b57a5e","target":"record","created_at":"2026-05-21T01:05:31Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"2bf6e8b15b7d24b329c9fde9c61f0033e9a84f94e4b293723ed84bb72932d814","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-20T10:33:11Z","title_canon_sha256":"bd5135c0a3f26fa9c6cfe4cf850bd019a5809642d5f4c42f3f70055ebf521dd5"},"schema_version":"1.0","source":{"id":"2605.20994","kind":"arxiv","version":1}},"canonical_sha256":"b8a9cc3cce74772aa376a1faa9d3db5ae659eaee07e5417bff487800d30f8c8a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b8a9cc3cce74772aa376a1faa9d3db5ae659eaee07e5417bff487800d30f8c8a","first_computed_at":"2026-05-21T01:05:31.576289Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-21T01:05:31.576289Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"anKdEI1oh3J/ULkXPwO4H15DhmVDqWK4QB6dzzHMl8ih0yC/OtMt2l8ABygjtPBmnL446OThMxHQWOcMGZhmAw==","signature_status":"signed_v1","signed_at":"2026-05-21T01:05:31.577012Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.20994","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:80f969be053e8bf862e5f3ce9feefebac59c8fd85135853794eb07db87b57a5e","sha256:6cfa3abc776091bb21ae4b3141983e635ebbada40c89de96efc02aa1cff2285a"],"state_sha256":"8f608b4a89592b2019a16531c8faf63cb0c09cfb044123d5bb78aa1079e1a250"}