{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MWQGWLLBL6UXJIKFYOPGLRCV6F","short_pith_number":"pith:MWQGWLLB","schema_version":"1.0","canonical_sha256":"65a06b2d615fa974a145c39e65c455f173e7a6299e31cfd80c4fd5ca22d5ea32","source":{"kind":"arxiv","id":"2406.18682","version":2},"attestation_state":"computed","paper":{"title":"The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Aakanksha, Arash Ahmadian, Beyza Ermis, Julia Kreutzer, Marzieh Fadaee, Sara Hooker, Seraphina Goldfarb-Tarrant","submitted_at":"2024-06-26T18:39:08Z","abstract_excerpt":"A key concern with the concept of \"alignment\" is the implicit question of \"alignment to what?\". AI systems are increasingly used across the world, yet safety alignment is often focused on homogeneous monolingual settings. Additionally, preference training and safety measures often overfit to harms common in Western-centric datasets. Here, we explore the viability of different alignment approaches when balancing dual objectives: addressing and optimizing for a non-homogeneous set of languages and cultural preferences while minimizing both global and local harms. We collect the first set of huma"},"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":"2406.18682","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-26T18:39:08Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"7e8d62b9fd3b73d50a47c5119b2d7530fa1b224b466d7c3923ee1530b55aa94b","abstract_canon_sha256":"d32300e3cc41283e75f60509bf24418cde506689d9ed6168690ddf40ed189a25"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:41:04.700214Z","signature_b64":"EEjrIiEd5mNBgyMmH06ghHaiwq12cBc0JcJ35oYlgQFJeL2kYQ64ZPPHAKLDoM/fzpYV8SCYqROqWRhGJCD+BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"65a06b2d615fa974a145c39e65c455f173e7a6299e31cfd80c4fd5ca22d5ea32","last_reissued_at":"2026-07-05T08:41:04.699629Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:41:04.699629Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Aakanksha, Arash Ahmadian, Beyza Ermis, Julia Kreutzer, Marzieh Fadaee, Sara Hooker, Seraphina Goldfarb-Tarrant","submitted_at":"2024-06-26T18:39:08Z","abstract_excerpt":"A key concern with the concept of \"alignment\" is the implicit question of \"alignment to what?\". AI systems are increasingly used across the world, yet safety alignment is often focused on homogeneous monolingual settings. Additionally, preference training and safety measures often overfit to harms common in Western-centric datasets. Here, we explore the viability of different alignment approaches when balancing dual objectives: addressing and optimizing for a non-homogeneous set of languages and cultural preferences while minimizing both global and local harms. We collect the first set of huma"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.18682","kind":"arxiv","version":2},"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/2406.18682/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":"2406.18682","created_at":"2026-07-05T08:41:04.699688+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.18682v2","created_at":"2026-07-05T08:41:04.699688+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.18682","created_at":"2026-07-05T08:41:04.699688+00:00"},{"alias_kind":"pith_short_12","alias_value":"MWQGWLLBL6UX","created_at":"2026-07-05T08:41:04.699688+00:00"},{"alias_kind":"pith_short_16","alias_value":"MWQGWLLBL6UXJIKF","created_at":"2026-07-05T08:41:04.699688+00:00"},{"alias_kind":"pith_short_8","alias_value":"MWQGWLLB","created_at":"2026-07-05T08:41:04.699688+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07874","citing_title":"Safety is Contextual, LLM-Judges Are Not: Navigating the Rigid Priors of Evaluators","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2510.21285","citing_title":"When Models Outthink Their Safety: Unveiling and Mitigating Self-Jailbreak in Large Reasoning Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14152","citing_title":"ROK-FORTRESS: Measuring the Effect of Geopolitical Transcreation for National Security and Public Safety","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14152","citing_title":"ROK-FORTRESS: Measuring the Effect of Geopolitical Transcreation for National Security and Public Safety","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05662","citing_title":"XL-SafetyBench: A Country-Grounded Cross-Cultural Benchmark for LLM Safety and Cultural Sensitivity","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F","json":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F.json","graph_json":"https://pith.science/api/pith-number/MWQGWLLBL6UXJIKFYOPGLRCV6F/graph.json","events_json":"https://pith.science/api/pith-number/MWQGWLLBL6UXJIKFYOPGLRCV6F/events.json","paper":"https://pith.science/paper/MWQGWLLB"},"agent_actions":{"view_html":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F","download_json":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F.json","view_paper":"https://pith.science/paper/MWQGWLLB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.18682&json=true","fetch_graph":"https://pith.science/api/pith-number/MWQGWLLBL6UXJIKFYOPGLRCV6F/graph.json","fetch_events":"https://pith.science/api/pith-number/MWQGWLLBL6UXJIKFYOPGLRCV6F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F/action/storage_attestation","attest_author":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F/action/author_attestation","sign_citation":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F/action/citation_signature","submit_replication":"https://pith.science/pith/MWQGWLLBL6UXJIKFYOPGLRCV6F/action/replication_record"}},"created_at":"2026-07-05T08:41:04.699688+00:00","updated_at":"2026-07-05T08:41:04.699688+00:00"}