{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YN4LFUT73KKYVW4S5FLDPZUT7S","short_pith_number":"pith:YN4LFUT7","schema_version":"1.0","canonical_sha256":"c378b2d27fda958adb92e95637e693fc8c29722c25c270ed924d78526f3a9b55","source":{"kind":"arxiv","id":"2607.07918","version":1},"attestation_state":"computed","paper":{"title":"Efficient Safety Alignment of Language Models via Latent Personality Traits","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Adam Oberman, Damiano Fornasiere, David Williams-King, Linh Le, Mohamed Amine Merzouk, Nolan Smyth","submitted_at":"2026-07-08T21:03:27Z","abstract_excerpt":"Current safety methods for large language models are known to be vulnerable to adversarial attacks, motivating research into robust alternatives. Latent Adversarial Training (LAT) is among the most effective defenses, but can degrade utility and requires training on large datasets of harmful prompts. We introduce Latent Personality Alignment (LPA), which replaces explicit harm refusal with adversarial training on just 66 harm-agnostic statements drawn from psychometric personality literature. We hypothesize that personality-anchored representations share latent structure with harm avoidance, s"},"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":"2607.07918","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-07-08T21:03:27Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CR"],"title_canon_sha256":"e6ab3f72b63ae2d03a16d8f0709c6bffda73bd28f378ab13f70b172c105cafd5","abstract_canon_sha256":"ab8df000db57eac304f08bea2c26d37a3352995b9461c3e4bd52c32ba0bbe0d2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-10T00:19:04.814915Z","signature_b64":"5jm2QNJbsZNQvu8qJqPXvyGa05LaN/7w561gcachmcedgDu9JGmi61tA1gobTHajhy9ZQjtCSb9W8+ow2SvwDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c378b2d27fda958adb92e95637e693fc8c29722c25c270ed924d78526f3a9b55","last_reissued_at":"2026-07-10T00:19:04.814441Z","signature_status":"signed_v1","first_computed_at":"2026-07-10T00:19:04.814441Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Efficient Safety Alignment of Language Models via Latent Personality Traits","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Adam Oberman, Damiano Fornasiere, David Williams-King, Linh Le, Mohamed Amine Merzouk, Nolan Smyth","submitted_at":"2026-07-08T21:03:27Z","abstract_excerpt":"Current safety methods for large language models are known to be vulnerable to adversarial attacks, motivating research into robust alternatives. Latent Adversarial Training (LAT) is among the most effective defenses, but can degrade utility and requires training on large datasets of harmful prompts. We introduce Latent Personality Alignment (LPA), which replaces explicit harm refusal with adversarial training on just 66 harm-agnostic statements drawn from psychometric personality literature. We hypothesize that personality-anchored representations share latent structure with harm avoidance, s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.07918","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/2607.07918/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":"2607.07918","created_at":"2026-07-10T00:19:04.814511+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.07918v1","created_at":"2026-07-10T00:19:04.814511+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.07918","created_at":"2026-07-10T00:19:04.814511+00:00"},{"alias_kind":"pith_short_12","alias_value":"YN4LFUT73KKY","created_at":"2026-07-10T00:19:04.814511+00:00"},{"alias_kind":"pith_short_16","alias_value":"YN4LFUT73KKYVW4S","created_at":"2026-07-10T00:19:04.814511+00:00"},{"alias_kind":"pith_short_8","alias_value":"YN4LFUT7","created_at":"2026-07-10T00:19:04.814511+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/YN4LFUT73KKYVW4S5FLDPZUT7S","json":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S.json","graph_json":"https://pith.science/api/pith-number/YN4LFUT73KKYVW4S5FLDPZUT7S/graph.json","events_json":"https://pith.science/api/pith-number/YN4LFUT73KKYVW4S5FLDPZUT7S/events.json","paper":"https://pith.science/paper/YN4LFUT7"},"agent_actions":{"view_html":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S","download_json":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S.json","view_paper":"https://pith.science/paper/YN4LFUT7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.07918&json=true","fetch_graph":"https://pith.science/api/pith-number/YN4LFUT73KKYVW4S5FLDPZUT7S/graph.json","fetch_events":"https://pith.science/api/pith-number/YN4LFUT73KKYVW4S5FLDPZUT7S/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S/action/storage_attestation","attest_author":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S/action/author_attestation","sign_citation":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S/action/citation_signature","submit_replication":"https://pith.science/pith/YN4LFUT73KKYVW4S5FLDPZUT7S/action/replication_record"}},"created_at":"2026-07-10T00:19:04.814511+00:00","updated_at":"2026-07-10T00:19:04.814511+00:00"}