{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:Z5Y4SNF4U7USJE4HTBTDHMHOGU","short_pith_number":"pith:Z5Y4SNF4","canonical_record":{"source":{"id":"2308.06463","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-12T04:05:57Z","cross_cats_sorted":[],"title_canon_sha256":"e1562e83b5c6eea3514aa0ccf0f4e20d7f549da5ae78057398c1cbe592c47183","abstract_canon_sha256":"683f998848559444496ea3f1ed6a7687717c67e3027f19c6aa570a6fe78cdb91"},"schema_version":"1.0"},"canonical_sha256":"cf71c934bca7e9249387986633b0ee350a1c0b59062bc14057e2a1a90da3acff","source":{"kind":"arxiv","id":"2308.06463","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2308.06463","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"arxiv_version","alias_value":"2308.06463v2","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.06463","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"pith_short_12","alias_value":"Z5Y4SNF4U7US","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"pith_short_16","alias_value":"Z5Y4SNF4U7USJE4H","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"pith_short_8","alias_value":"Z5Y4SNF4","created_at":"2026-07-05T08:00:34Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:Z5Y4SNF4U7USJE4HTBTDHMHOGU","target":"record","payload":{"canonical_record":{"source":{"id":"2308.06463","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-12T04:05:57Z","cross_cats_sorted":[],"title_canon_sha256":"e1562e83b5c6eea3514aa0ccf0f4e20d7f549da5ae78057398c1cbe592c47183","abstract_canon_sha256":"683f998848559444496ea3f1ed6a7687717c67e3027f19c6aa570a6fe78cdb91"},"schema_version":"1.0"},"canonical_sha256":"cf71c934bca7e9249387986633b0ee350a1c0b59062bc14057e2a1a90da3acff","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:00:34.997744Z","signature_b64":"NjWXybxavztm+8S9afPRo4Hple2dcLSyuecBdBpXpObSkJ5g0AWE5RTWSbnR2oNT8D4sNbre8abbv4LLC9N8Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cf71c934bca7e9249387986633b0ee350a1c0b59062bc14057e2a1a90da3acff","last_reissued_at":"2026-07-05T08:00:34.997156Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:00:34.997156Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2308.06463","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:00:34Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BYdF2zEvAsy6a3XHcerNg22QLGFHy18dEpCAzNWrnQNNOJLV9Sqe5mGmCAkRPTY8w9Dd1CcyXEvwOE8wMP+4Aw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-13T15:11:18.459524Z"},"content_sha256":"18b04953c7c96122214a0341dbc7b2727e30e1f6b308180da75853c428702a18","schema_version":"1.0","event_id":"sha256:18b04953c7c96122214a0341dbc7b2727e30e1f6b308180da75853c428702a18"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:Z5Y4SNF4U7USJE4HTBTDHMHOGU","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jen-tse Huang, Pinjia He, Shuming Shi, Wenxiang Jiao, Wenxuan Wang, Youliang Yuan, Zhaopeng Tu","submitted_at":"2023-08-12T04:05:57Z","abstract_excerpt":"Safety lies at the core of the development of Large Language Models (LLMs). There is ample work on aligning LLMs with human ethics and preferences, including data filtering in pretraining, supervised fine-tuning, reinforcement learning from human feedback, and red teaming, etc. In this study, we discover that chat in cipher can bypass the safety alignment techniques of LLMs, which are mainly conducted in natural languages. We propose a novel framework CipherChat to systematically examine the generalizability of safety alignment to non-natural languages -- ciphers. CipherChat enables humans to "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.06463","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/2308.06463/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":"2026-07-05T08:00:34Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mkup0NCv5whvnYL/FLZt0UMed/29QIsrLtkOfGL0E+c6WDSxqLkj8g2VS6wAYBwiOSpz4/ZmYzHvtbYyGic2Dw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-13T15:11:18.459893Z"},"content_sha256":"84eed013c484991925477b3a74ddf2eebd2dc6a099e618bef67f0cf480499c6c","schema_version":"1.0","event_id":"sha256:84eed013c484991925477b3a74ddf2eebd2dc6a099e618bef67f0cf480499c6c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/Z5Y4SNF4U7USJE4HTBTDHMHOGU/bundle.json","state_url":"https://pith.science/pith/Z5Y4SNF4U7USJE4HTBTDHMHOGU/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/Z5Y4SNF4U7USJE4HTBTDHMHOGU/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":"2026-07-13T15:11:18Z","links":{"resolver":"https://pith.science/pith/Z5Y4SNF4U7USJE4HTBTDHMHOGU","bundle":"https://pith.science/pith/Z5Y4SNF4U7USJE4HTBTDHMHOGU/bundle.json","state":"https://pith.science/pith/Z5Y4SNF4U7USJE4HTBTDHMHOGU/state.json","well_known_bundle":"https://pith.science/.well-known/pith/Z5Y4SNF4U7USJE4HTBTDHMHOGU/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:Z5Y4SNF4U7USJE4HTBTDHMHOGU","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":"683f998848559444496ea3f1ed6a7687717c67e3027f19c6aa570a6fe78cdb91","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-12T04:05:57Z","title_canon_sha256":"e1562e83b5c6eea3514aa0ccf0f4e20d7f549da5ae78057398c1cbe592c47183"},"schema_version":"1.0","source":{"id":"2308.06463","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2308.06463","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"arxiv_version","alias_value":"2308.06463v2","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.06463","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"pith_short_12","alias_value":"Z5Y4SNF4U7US","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"pith_short_16","alias_value":"Z5Y4SNF4U7USJE4H","created_at":"2026-07-05T08:00:34Z"},{"alias_kind":"pith_short_8","alias_value":"Z5Y4SNF4","created_at":"2026-07-05T08:00:34Z"}],"graph_snapshots":[{"event_id":"sha256:84eed013c484991925477b3a74ddf2eebd2dc6a099e618bef67f0cf480499c6c","target":"graph","created_at":"2026-07-05T08:00:34Z","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/2308.06463/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Safety lies at the core of the development of Large Language Models (LLMs). There is ample work on aligning LLMs with human ethics and preferences, including data filtering in pretraining, supervised fine-tuning, reinforcement learning from human feedback, and red teaming, etc. In this study, we discover that chat in cipher can bypass the safety alignment techniques of LLMs, which are mainly conducted in natural languages. We propose a novel framework CipherChat to systematically examine the generalizability of safety alignment to non-natural languages -- ciphers. CipherChat enables humans to ","authors_text":"Jen-tse Huang, Pinjia He, Shuming Shi, Wenxiang Jiao, Wenxuan Wang, Youliang Yuan, Zhaopeng Tu","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-12T04:05:57Z","title":"GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.06463","kind":"arxiv","version":2},"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":"sha256:18b04953c7c96122214a0341dbc7b2727e30e1f6b308180da75853c428702a18","target":"record","created_at":"2026-07-05T08:00:34Z","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":"683f998848559444496ea3f1ed6a7687717c67e3027f19c6aa570a6fe78cdb91","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-12T04:05:57Z","title_canon_sha256":"e1562e83b5c6eea3514aa0ccf0f4e20d7f549da5ae78057398c1cbe592c47183"},"schema_version":"1.0","source":{"id":"2308.06463","kind":"arxiv","version":2}},"canonical_sha256":"cf71c934bca7e9249387986633b0ee350a1c0b59062bc14057e2a1a90da3acff","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"cf71c934bca7e9249387986633b0ee350a1c0b59062bc14057e2a1a90da3acff","first_computed_at":"2026-07-05T08:00:34.997156Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:00:34.997156Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"NjWXybxavztm+8S9afPRo4Hple2dcLSyuecBdBpXpObSkJ5g0AWE5RTWSbnR2oNT8D4sNbre8abbv4LLC9N8Cg==","signature_status":"signed_v1","signed_at":"2026-07-05T08:00:34.997744Z","signed_message":"canonical_sha256_bytes"},"source_id":"2308.06463","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:18b04953c7c96122214a0341dbc7b2727e30e1f6b308180da75853c428702a18","sha256:84eed013c484991925477b3a74ddf2eebd2dc6a099e618bef67f0cf480499c6c"],"state_sha256":"e183bfc2df544345788ce82902ba067bde082fa924ef0d99d111e6a4f030b064"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hK4rrRUD3HXOFzP7VyjkmbVv2bw6IIaTr9LRSKI9TD3iaRg75FrgE5lTT3dcOhdmCbrrztBgTT6HliWmM6YrCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-13T15:11:18.461943Z","bundle_sha256":"322e5d3ce7d3a04adbb9fcd484dd6b65a82762bc0ea876b26092ffc47c8342ea"}}