{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:4BJL7TO6FQH2UL7G2MC5JEEPRC","short_pith_number":"pith:4BJL7TO6","schema_version":"1.0","canonical_sha256":"e052bfcdde2c0faa2fe6d305d4908f88a261ded63154ed28e7e7f842db94119e","source":{"kind":"arxiv","id":"2607.09053","version":1},"attestation_state":"computed","paper":{"title":"An Emergent Mirage: Is Emergent Misalignment and Realignment Indeed a Robust Phenomenon?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Abhinav Rao, Atharva Naik, Bin Hu, Liancheng Gong","submitted_at":"2026-07-10T02:50:21Z","abstract_excerpt":"Recent work has reported Emergent Misalignment (EM), where language models fine-tuned on narrow, domain-specific misaligned datasets abruptly acquire broadly misaligned behavior, alongside evidence that this behavior can be reversed through limited realignment. We systematically study repeated alignment and misalignment cycles using controlled fine-tuning loops while tracking behavioral performance, and LoRA representations throughout training. Although we reproduce EM, we find that both misalignment and realignment are highly sensitive to superficial dataset characteristics, with apparent rap"},"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":"2607.09053","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-10T02:50:21Z","cross_cats_sorted":[],"title_canon_sha256":"b9930ad61bcdba7c182d121dee22164e72c48da786d032ac3d12616f32b6f234","abstract_canon_sha256":"ad614c1d9f4c883cbbb06010c0438212ea98d6e4b35312db8f77870ba0250391"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-13T00:17:37.989226Z","signature_b64":"486Pn+XkCfqb7RB4fdWo3MR2D8r0JM7OdjvRIgT7dCN4Emv5LE/rS9PgPahXM3P1DC5P/YrkhlNtYMBLjRmPAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e052bfcdde2c0faa2fe6d305d4908f88a261ded63154ed28e7e7f842db94119e","last_reissued_at":"2026-07-13T00:17:37.988197Z","signature_status":"signed_v1","first_computed_at":"2026-07-13T00:17:37.988197Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"An Emergent Mirage: Is Emergent Misalignment and Realignment Indeed a Robust Phenomenon?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Abhinav Rao, Atharva Naik, Bin Hu, Liancheng Gong","submitted_at":"2026-07-10T02:50:21Z","abstract_excerpt":"Recent work has reported Emergent Misalignment (EM), where language models fine-tuned on narrow, domain-specific misaligned datasets abruptly acquire broadly misaligned behavior, alongside evidence that this behavior can be reversed through limited realignment. We systematically study repeated alignment and misalignment cycles using controlled fine-tuning loops while tracking behavioral performance, and LoRA representations throughout training. Although we reproduce EM, we find that both misalignment and realignment are highly sensitive to superficial dataset characteristics, with apparent rap"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.09053","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/2607.09053/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":"2607.09053","created_at":"2026-07-13T00:17:37.988717+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.09053v1","created_at":"2026-07-13T00:17:37.988717+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.09053","created_at":"2026-07-13T00:17:37.988717+00:00"},{"alias_kind":"pith_short_12","alias_value":"4BJL7TO6FQH2","created_at":"2026-07-13T00:17:37.988717+00:00"},{"alias_kind":"pith_short_16","alias_value":"4BJL7TO6FQH2UL7G","created_at":"2026-07-13T00:17:37.988717+00:00"},{"alias_kind":"pith_short_8","alias_value":"4BJL7TO6","created_at":"2026-07-13T00:17:37.988717+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/4BJL7TO6FQH2UL7G2MC5JEEPRC","json":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC.json","graph_json":"https://pith.science/api/pith-number/4BJL7TO6FQH2UL7G2MC5JEEPRC/graph.json","events_json":"https://pith.science/api/pith-number/4BJL7TO6FQH2UL7G2MC5JEEPRC/events.json","paper":"https://pith.science/paper/4BJL7TO6"},"agent_actions":{"view_html":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC","download_json":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC.json","view_paper":"https://pith.science/paper/4BJL7TO6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.09053&json=true","fetch_graph":"https://pith.science/api/pith-number/4BJL7TO6FQH2UL7G2MC5JEEPRC/graph.json","fetch_events":"https://pith.science/api/pith-number/4BJL7TO6FQH2UL7G2MC5JEEPRC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC/action/storage_attestation","attest_author":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC/action/author_attestation","sign_citation":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC/action/citation_signature","submit_replication":"https://pith.science/pith/4BJL7TO6FQH2UL7G2MC5JEEPRC/action/replication_record"}},"created_at":"2026-07-13T00:17:37.988717+00:00","updated_at":"2026-07-13T00:17:37.988717+00:00"}