{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:TSOVKV6GF2EKIXXGXB7VQHGTE4","short_pith_number":"pith:TSOVKV6G","canonical_record":{"source":{"id":"2505.16869","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-22T16:24:51Z","cross_cats_sorted":[],"title_canon_sha256":"d6066210e047785ff958b46710ac53c7e0bd1e1551aa1239ce2fbb3ef324f044","abstract_canon_sha256":"f6e833922e8c6a64e880b555527e87cf63245ca72f124473419f34f5d5aa3bf2"},"schema_version":"1.0"},"canonical_sha256":"9c9d5557c62e88a45ee6b87f581cd3270b3272d3d151ece75cae4fafc324f768","source":{"kind":"arxiv","id":"2505.16869","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.16869","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"arxiv_version","alias_value":"2505.16869v1","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.16869","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"pith_short_12","alias_value":"TSOVKV6GF2EK","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"pith_short_16","alias_value":"TSOVKV6GF2EKIXXG","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"pith_short_8","alias_value":"TSOVKV6G","created_at":"2026-07-05T11:07:42Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:TSOVKV6GF2EKIXXGXB7VQHGTE4","target":"record","payload":{"canonical_record":{"source":{"id":"2505.16869","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-22T16:24:51Z","cross_cats_sorted":[],"title_canon_sha256":"d6066210e047785ff958b46710ac53c7e0bd1e1551aa1239ce2fbb3ef324f044","abstract_canon_sha256":"f6e833922e8c6a64e880b555527e87cf63245ca72f124473419f34f5d5aa3bf2"},"schema_version":"1.0"},"canonical_sha256":"9c9d5557c62e88a45ee6b87f581cd3270b3272d3d151ece75cae4fafc324f768","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:07:42.621906Z","signature_b64":"1D2Wm8m3s6I5FcxqNs/XuxZn0LOvDKuRla5fqdlHGanuDxhLLKU+pQofwKEDGr2exqz8JK4HSveAO/Mffe25Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9c9d5557c62e88a45ee6b87f581cd3270b3272d3d151ece75cae4fafc324f768","last_reissued_at":"2026-07-05T11:07:42.621372Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:07:42.621372Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2505.16869","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:07:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wQkeouKgJCxlu5zbqGh2T4Ec92DnWKGzHsWlYL49V4ofxrWL83oiiK3n0pT5J/Q+yRQz2LJkP/ROjuaaSxjnAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-18T06:08:03.465902Z"},"content_sha256":"4250fa1dca1905f6a888f9b0324bf799feb497713ecd1c2668514cdfb0a6c0ea","schema_version":"1.0","event_id":"sha256:4250fa1dca1905f6a888f9b0324bf799feb497713ecd1c2668514cdfb0a6c0ea"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:TSOVKV6GF2EKIXXGXB7VQHGTE4","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"MPO: Multilingual Safety Alignment via Reward Gap Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"An Zhang, Bing Qin, Jiahe Guo, Tat-Seng Chua, Ting Liu, Tongtong Wu, Weixiang Zhao, Wenxuan Zhang, Yang Deng, Yanyan Zhao, Yulin Hu","submitted_at":"2025-05-22T16:24:51Z","abstract_excerpt":"Large language models (LLMs) have become increasingly central to AI applications worldwide, necessitating robust multilingual safety alignment to ensure secure deployment across diverse linguistic contexts. Existing preference learning methods for safety alignment, such as RLHF and DPO, are primarily monolingual and struggle with noisy multilingual data. To address these limitations, we introduce Multilingual reward gaP Optimization (MPO), a novel approach that leverages the well-aligned safety capabilities of the dominant language (English) to improve safety alignment across multiple language"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.16869","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/2505.16869/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:07:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"M5nTuVZEw0LuFsssW126a0fwjXMiybhpH9LP0dNLGi3wfQlBRMLufQUIjIux/K2kYZcjeCLhaFkzSl3Yj3vaCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-18T06:08:03.466423Z"},"content_sha256":"1fb54422f1173233187680e2c43118b1e7aaeaf3aec4220d20177c142499cf22","schema_version":"1.0","event_id":"sha256:1fb54422f1173233187680e2c43118b1e7aaeaf3aec4220d20177c142499cf22"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TSOVKV6GF2EKIXXGXB7VQHGTE4/bundle.json","state_url":"https://pith.science/pith/TSOVKV6GF2EKIXXGXB7VQHGTE4/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TSOVKV6GF2EKIXXGXB7VQHGTE4/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-18T06:08:03Z","links":{"resolver":"https://pith.science/pith/TSOVKV6GF2EKIXXGXB7VQHGTE4","bundle":"https://pith.science/pith/TSOVKV6GF2EKIXXGXB7VQHGTE4/bundle.json","state":"https://pith.science/pith/TSOVKV6GF2EKIXXGXB7VQHGTE4/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TSOVKV6GF2EKIXXGXB7VQHGTE4/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:TSOVKV6GF2EKIXXGXB7VQHGTE4","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":"f6e833922e8c6a64e880b555527e87cf63245ca72f124473419f34f5d5aa3bf2","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-22T16:24:51Z","title_canon_sha256":"d6066210e047785ff958b46710ac53c7e0bd1e1551aa1239ce2fbb3ef324f044"},"schema_version":"1.0","source":{"id":"2505.16869","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.16869","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"arxiv_version","alias_value":"2505.16869v1","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.16869","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"pith_short_12","alias_value":"TSOVKV6GF2EK","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"pith_short_16","alias_value":"TSOVKV6GF2EKIXXG","created_at":"2026-07-05T11:07:42Z"},{"alias_kind":"pith_short_8","alias_value":"TSOVKV6G","created_at":"2026-07-05T11:07:42Z"}],"graph_snapshots":[{"event_id":"sha256:1fb54422f1173233187680e2c43118b1e7aaeaf3aec4220d20177c142499cf22","target":"graph","created_at":"2026-07-05T11:07:42Z","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/2505.16869/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) have become increasingly central to AI applications worldwide, necessitating robust multilingual safety alignment to ensure secure deployment across diverse linguistic contexts. Existing preference learning methods for safety alignment, such as RLHF and DPO, are primarily monolingual and struggle with noisy multilingual data. To address these limitations, we introduce Multilingual reward gaP Optimization (MPO), a novel approach that leverages the well-aligned safety capabilities of the dominant language (English) to improve safety alignment across multiple language","authors_text":"An Zhang, Bing Qin, Jiahe Guo, Tat-Seng Chua, Ting Liu, Tongtong Wu, Weixiang Zhao, Wenxuan Zhang, Yang Deng, Yanyan Zhao, Yulin Hu","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-22T16:24:51Z","title":"MPO: Multilingual Safety Alignment via Reward Gap Optimization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.16869","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":"sha256:4250fa1dca1905f6a888f9b0324bf799feb497713ecd1c2668514cdfb0a6c0ea","target":"record","created_at":"2026-07-05T11:07:42Z","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":"f6e833922e8c6a64e880b555527e87cf63245ca72f124473419f34f5d5aa3bf2","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-22T16:24:51Z","title_canon_sha256":"d6066210e047785ff958b46710ac53c7e0bd1e1551aa1239ce2fbb3ef324f044"},"schema_version":"1.0","source":{"id":"2505.16869","kind":"arxiv","version":1}},"canonical_sha256":"9c9d5557c62e88a45ee6b87f581cd3270b3272d3d151ece75cae4fafc324f768","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9c9d5557c62e88a45ee6b87f581cd3270b3272d3d151ece75cae4fafc324f768","first_computed_at":"2026-07-05T11:07:42.621372Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:07:42.621372Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"1D2Wm8m3s6I5FcxqNs/XuxZn0LOvDKuRla5fqdlHGanuDxhLLKU+pQofwKEDGr2exqz8JK4HSveAO/Mffe25Cg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:07:42.621906Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.16869","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4250fa1dca1905f6a888f9b0324bf799feb497713ecd1c2668514cdfb0a6c0ea","sha256:1fb54422f1173233187680e2c43118b1e7aaeaf3aec4220d20177c142499cf22"],"state_sha256":"fa3f13b9d0de487f8a8700739472b9d0877c1f20ed2d25577ab589155c384141"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"X7DH6DHCQmhe1fw7nz6JAk1T2669KJeWXQPvu+CkTTSb54sjilHWiBuV0/byZXSryV+9Yd8rzdbVqxbiEBcXBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-18T06:08:03.470322Z","bundle_sha256":"92e94368d42f1735b2b68848f97528ae77d6e899102e5fd5315e52906d267303"}}