{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:IKPBBXOVWHXF5475IXXOEJQMZG","short_pith_number":"pith:IKPBBXOV","canonical_record":{"source":{"id":"2304.12298","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-02-21T08:59:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7eddf8e4bc78a0783dd3f3895a754b54cced5901f22ca7022eb7de0a00b9aaad","abstract_canon_sha256":"704f7ba4f2881626139008db563f9d7dd0627b7ea099cb2eb02b19df12e5ea84"},"schema_version":"1.0"},"canonical_sha256":"429e10ddd5b1ee5ef3fd45eee2260cc9bb32ca46e5500618778e910736eedb2d","source":{"kind":"arxiv","id":"2304.12298","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2304.12298","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"arxiv_version","alias_value":"2304.12298v1","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.12298","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"pith_short_12","alias_value":"IKPBBXOVWHXF","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"pith_short_16","alias_value":"IKPBBXOVWHXF5475","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"pith_short_8","alias_value":"IKPBBXOV","created_at":"2026-07-05T06:03:50Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:IKPBBXOVWHXF5475IXXOEJQMZG","target":"record","payload":{"canonical_record":{"source":{"id":"2304.12298","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-02-21T08:59:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7eddf8e4bc78a0783dd3f3895a754b54cced5901f22ca7022eb7de0a00b9aaad","abstract_canon_sha256":"704f7ba4f2881626139008db563f9d7dd0627b7ea099cb2eb02b19df12e5ea84"},"schema_version":"1.0"},"canonical_sha256":"429e10ddd5b1ee5ef3fd45eee2260cc9bb32ca46e5500618778e910736eedb2d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:03:50.077524Z","signature_b64":"v6zrS92jiogfeMNKHmaNjtwosChSJ4dWZSRtqpsggXs0gWYcLTi26RxngLgXcs28yMWdCpDUBxttyGuagId5CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"429e10ddd5b1ee5ef3fd45eee2260cc9bb32ca46e5500618778e910736eedb2d","last_reissued_at":"2026-07-05T06:03:50.077014Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:03:50.077014Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2304.12298","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":"2026-07-05T06:03:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zjPAhiZ6h8V8wWjFOFO8Q2JvqXAALTOb42tRazTPbJYty3SvnPoYHnMOBrHJ3pQNnICJ7URsCyOmH1Pd+MLKAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-16T20:39:37.521433Z"},"content_sha256":"a020643e0ca440b9c1654f6b43aa8652b214e0636daef0011c517b8a3daaf13e","schema_version":"1.0","event_id":"sha256:a020643e0ca440b9c1654f6b43aa8652b214e0636daef0011c517b8a3daaf13e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:IKPBBXOVWHXF5475IXXOEJQMZG","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"BadGPT: Exploring Security Vulnerabilities of ChatGPT via Backdoor Attacks to InstructGPT","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Jiawen Shi, Lichao Sun, Pan Zhou, Yixin Liu","submitted_at":"2023-02-21T08:59:22Z","abstract_excerpt":"Recently, ChatGPT has gained significant attention in research due to its ability to interact with humans effectively. The core idea behind this model is reinforcement learning (RL) fine-tuning, a new paradigm that allows language models to align with human preferences, i.e., InstructGPT. In this study, we propose BadGPT, the first backdoor attack against RL fine-tuning in language models. By injecting a backdoor into the reward model, the language model can be compromised during the fine-tuning stage. Our initial experiments on movie reviews, i.e., IMDB, demonstrate that an attacker can manip"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.12298","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/2304.12298/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":"2026-07-05T06:03:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7UXy9lnDzAX+YAQNIlTNhND+Ccuyue3+xQyQYhjg8rhVRcyLdTpGp/3P0y9t6kc5edEgIRhhNrFZubHAZkHPDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-16T20:39:37.521939Z"},"content_sha256":"4ad52e549d4222a05f14f1bd7a7c5d4355babc553633d5205c2fb9c18c6b14a5","schema_version":"1.0","event_id":"sha256:4ad52e549d4222a05f14f1bd7a7c5d4355babc553633d5205c2fb9c18c6b14a5"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IKPBBXOVWHXF5475IXXOEJQMZG/bundle.json","state_url":"https://pith.science/pith/IKPBBXOVWHXF5475IXXOEJQMZG/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IKPBBXOVWHXF5475IXXOEJQMZG/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":"2026-08-16T20:39:37Z","links":{"resolver":"https://pith.science/pith/IKPBBXOVWHXF5475IXXOEJQMZG","bundle":"https://pith.science/pith/IKPBBXOVWHXF5475IXXOEJQMZG/bundle.json","state":"https://pith.science/pith/IKPBBXOVWHXF5475IXXOEJQMZG/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IKPBBXOVWHXF5475IXXOEJQMZG/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:IKPBBXOVWHXF5475IXXOEJQMZG","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":"704f7ba4f2881626139008db563f9d7dd0627b7ea099cb2eb02b19df12e5ea84","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-02-21T08:59:22Z","title_canon_sha256":"7eddf8e4bc78a0783dd3f3895a754b54cced5901f22ca7022eb7de0a00b9aaad"},"schema_version":"1.0","source":{"id":"2304.12298","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2304.12298","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"arxiv_version","alias_value":"2304.12298v1","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.12298","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"pith_short_12","alias_value":"IKPBBXOVWHXF","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"pith_short_16","alias_value":"IKPBBXOVWHXF5475","created_at":"2026-07-05T06:03:50Z"},{"alias_kind":"pith_short_8","alias_value":"IKPBBXOV","created_at":"2026-07-05T06:03:50Z"}],"graph_snapshots":[{"event_id":"sha256:4ad52e549d4222a05f14f1bd7a7c5d4355babc553633d5205c2fb9c18c6b14a5","target":"graph","created_at":"2026-07-05T06:03:50Z","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/2304.12298/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recently, ChatGPT has gained significant attention in research due to its ability to interact with humans effectively. The core idea behind this model is reinforcement learning (RL) fine-tuning, a new paradigm that allows language models to align with human preferences, i.e., InstructGPT. In this study, we propose BadGPT, the first backdoor attack against RL fine-tuning in language models. By injecting a backdoor into the reward model, the language model can be compromised during the fine-tuning stage. Our initial experiments on movie reviews, i.e., IMDB, demonstrate that an attacker can manip","authors_text":"Jiawen Shi, Lichao Sun, Pan Zhou, Yixin Liu","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-02-21T08:59:22Z","title":"BadGPT: Exploring Security Vulnerabilities of ChatGPT via Backdoor Attacks to InstructGPT"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.12298","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:a020643e0ca440b9c1654f6b43aa8652b214e0636daef0011c517b8a3daaf13e","target":"record","created_at":"2026-07-05T06:03:50Z","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":"704f7ba4f2881626139008db563f9d7dd0627b7ea099cb2eb02b19df12e5ea84","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-02-21T08:59:22Z","title_canon_sha256":"7eddf8e4bc78a0783dd3f3895a754b54cced5901f22ca7022eb7de0a00b9aaad"},"schema_version":"1.0","source":{"id":"2304.12298","kind":"arxiv","version":1}},"canonical_sha256":"429e10ddd5b1ee5ef3fd45eee2260cc9bb32ca46e5500618778e910736eedb2d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"429e10ddd5b1ee5ef3fd45eee2260cc9bb32ca46e5500618778e910736eedb2d","first_computed_at":"2026-07-05T06:03:50.077014Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T06:03:50.077014Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"v6zrS92jiogfeMNKHmaNjtwosChSJ4dWZSRtqpsggXs0gWYcLTi26RxngLgXcs28yMWdCpDUBxttyGuagId5CA==","signature_status":"signed_v1","signed_at":"2026-07-05T06:03:50.077524Z","signed_message":"canonical_sha256_bytes"},"source_id":"2304.12298","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a020643e0ca440b9c1654f6b43aa8652b214e0636daef0011c517b8a3daaf13e","sha256:4ad52e549d4222a05f14f1bd7a7c5d4355babc553633d5205c2fb9c18c6b14a5"],"state_sha256":"eee45f5de3bcad0b24dcd0476d6da68b4b71f5f45b31552b20c891d5ca74b2c0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sgmZs8ZLUduhkMXMGLoWgvFfILxPDm9Lsdf+el+7CvoLccrsYzupJjqseU1iXJuhn4KxMCoMJUw2giZS03cwDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-16T20:39:37.527075Z","bundle_sha256":"6ff344ace5965f6e938e777baff176a40e6f35de81ea2c928db94af970e7a674"}}