{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:2C5FQKYPGW24CFMA4OGXYFHUNI","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":"30e80a9acc7c61f31bb9748fe5e9274c34733005765d8178bff8e059c3b3b223","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-03T21:30:56Z","title_canon_sha256":"33bbf057176543a71f37ad4685c2ce1ea6fbd47b59450069528b23f272988709"},"schema_version":"1.0","source":{"id":"2310.02446","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.02446","created_at":"2026-05-17T23:38:14Z"},{"alias_kind":"arxiv_version","alias_value":"2310.02446v2","created_at":"2026-05-17T23:38:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.02446","created_at":"2026-05-17T23:38:14Z"},{"alias_kind":"pith_short_12","alias_value":"2C5FQKYPGW24","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"2C5FQKYPGW24CFMA","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"2C5FQKYP","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:24e1b46473d15d2ce4c755e94c29642f8e6dde9e5c9f0ca8c7c2785c439706e7","target":"graph","created_at":"2026-05-17T23:38:14Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"On the AdvBenchmark, GPT-4 engages with the unsafe translated inputs and provides actionable items that can get the users towards their harmful goals 79% of the time, which is on par with or even surpassing state-of-the-art jailbreaking attacks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That automatic translations preserve the original unsafe intent without introducing detectable artifacts or triggering the model's cross-lingual safety mechanisms, and that results on the tested benchmark generalize to other prompts and models."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Translating unsafe inputs to low-resource languages jailbreaks GPT-4 at rates on par with or exceeding state-of-the-art attacks."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Translating harmful English prompts into low-resource languages lets GPT-4 provide actionable advice for bad goals 79 percent of the time."}],"snapshot_sha256":"5acc8099a82390bf1e378b69ad7a2679c90d48e86db7de89c166a8b17df892d1"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"685cae991da7393e3601bb01782359bffbed4eef16ae8255aba754e370d4704d"},"paper":{"abstract_excerpt":"AI safety training and red-teaming of large language models (LLMs) are measures to mitigate the generation of unsafe content. Our work exposes the inherent cross-lingual vulnerability of these safety mechanisms, resulting from the linguistic inequality of safety training data, by successfully circumventing GPT-4's safeguard through translating unsafe English inputs into low-resource languages. On the AdvBenchmark, GPT-4 engages with the unsafe translated inputs and provides actionable items that can get the users towards their harmful goals 79% of the time, which is on par with or even surpass","authors_text":"Cristina Menghini, Stephen H. Bach, Zheng-Xin Yong","cross_cats":["cs.AI","cs.CR","cs.LG"],"headline":"Translating harmful English prompts into low-resource languages lets GPT-4 provide actionable advice for bad goals 79 percent of the time.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-03T21:30:56Z","title":"Low-Resource Languages Jailbreak GPT-4"},"references":{"count":55,"internal_anchors":13,"resolved_work":55,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Jigsaw multilingual toxic comment classification, 2020","work_id":"acd2de50-2c59-482c-82e8-153db57bc84d","year":2020},{"cited_arxiv_id":"2204.05862","doi":"","is_internal_anchor":true,"ref_index":2,"title":"Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback","work_id":"a1f2574b-a899-4713-be60-c87ba332656c","year":2022},{"cited_arxiv_id":"2212.08073","doi":"","is_internal_anchor":true,"ref_index":3,"title":"Constitutional AI: Harmlessness from AI Feedback","work_id":"faaaa4e0-2676-4fac-a0b4-99aef10d2095","year":2022},{"cited_arxiv_id":"2302.04023","doi":"","is_internal_anchor":true,"ref_index":4,"title":"A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivity","work_id":"41dff74c-00b2-4c77-a674-9f86030c06c8","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"X., Macherey, K., Krikun, M., Wang, P., Gutkin, A., Shah, A., Huang, Y., Chen, Z., Wu, Y., and Hughes, M","work_id":"c55757ac-fd65-4628-8d13-7ce5b6c20ff0","year":2022}],"snapshot_sha256":"7c923c5d51f1b6baab3a00891cd1dffa7a58bf223887463a1b27ff32553e91a3"},"source":{"id":"2310.02446","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-17T09:12:43.135443Z","id":"fe7316e4-a497-4337-b3d2-1f0fc27eff95","model_set":{"reader":"grok-4.3"},"one_line_summary":"Translating unsafe inputs to low-resource languages jailbreaks GPT-4 at rates on par with or exceeding state-of-the-art attacks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Translating harmful English prompts into low-resource languages lets GPT-4 provide actionable advice for bad goals 79 percent of the time.","strongest_claim":"On the AdvBenchmark, GPT-4 engages with the unsafe translated inputs and provides actionable items that can get the users towards their harmful goals 79% of the time, which is on par with or even surpassing state-of-the-art jailbreaking attacks.","weakest_assumption":"That automatic translations preserve the original unsafe intent without introducing detectable artifacts or triggering the model's cross-lingual safety mechanisms, and that results on the tested benchmark generalize to other prompts and models."}},"verdict_id":"fe7316e4-a497-4337-b3d2-1f0fc27eff95"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:2814dfea75d6b7347c6a2f5ddb6922e822361ea058437e039f191353b911089d","target":"record","created_at":"2026-05-17T23:38:14Z","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":"30e80a9acc7c61f31bb9748fe5e9274c34733005765d8178bff8e059c3b3b223","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-03T21:30:56Z","title_canon_sha256":"33bbf057176543a71f37ad4685c2ce1ea6fbd47b59450069528b23f272988709"},"schema_version":"1.0","source":{"id":"2310.02446","kind":"arxiv","version":2}},"canonical_sha256":"d0ba582b0f35b5c11580e38d7c14f46a19141ba58e5252d5433334f7f24ce5cc","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d0ba582b0f35b5c11580e38d7c14f46a19141ba58e5252d5433334f7f24ce5cc","first_computed_at":"2026-05-17T23:38:14.486235Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:14.486235Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2h8Mb/uDfoxfPSBTjg5RX/HTGzB5mofwKXE4U2pHdxlloHjEhyNK3GwHFc1FpAHQRQCBHIu6DYBoyvATj9mTDw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:14.486936Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.02446","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:2814dfea75d6b7347c6a2f5ddb6922e822361ea058437e039f191353b911089d","sha256:24e1b46473d15d2ce4c755e94c29642f8e6dde9e5c9f0ca8c7c2785c439706e7"],"state_sha256":"a23cdfd0a25d0f9958fa038eb36e610c56ccc3991cb0b93cee276e023a666fd0"}