{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:JQ6CZIYLJZ2RGDLDBUZ6YAUIWE","short_pith_number":"pith:JQ6CZIYL","canonical_record":{"source":{"id":"2312.12321","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2023-12-19T16:47:12Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"9adf5facd573caed217df7a48ed5fc17f5651353e17e6082d9151fbc09ad6e9c","abstract_canon_sha256":"68aeea68b057f65d05c3def9444b44116a1ee7b8ac93ab8da0bc77ca78af3c60"},"schema_version":"1.0"},"canonical_sha256":"4c3c2ca30b4e75130d630d33ec0288b12cbca8e4c7a40e4326ca20cf7ebadad2","source":{"kind":"arxiv","id":"2312.12321","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.12321","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"arxiv_version","alias_value":"2312.12321v2","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.12321","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"pith_short_12","alias_value":"JQ6CZIYLJZ2R","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"pith_short_16","alias_value":"JQ6CZIYLJZ2RGDLD","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"pith_short_8","alias_value":"JQ6CZIYL","created_at":"2026-07-05T08:20:09Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:JQ6CZIYLJZ2RGDLDBUZ6YAUIWE","target":"record","payload":{"canonical_record":{"source":{"id":"2312.12321","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2023-12-19T16:47:12Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"9adf5facd573caed217df7a48ed5fc17f5651353e17e6082d9151fbc09ad6e9c","abstract_canon_sha256":"68aeea68b057f65d05c3def9444b44116a1ee7b8ac93ab8da0bc77ca78af3c60"},"schema_version":"1.0"},"canonical_sha256":"4c3c2ca30b4e75130d630d33ec0288b12cbca8e4c7a40e4326ca20cf7ebadad2","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:20:09.200835Z","signature_b64":"P8uuEqaIl3tqY8qah7aaJr4fbuTBFlsPCm/NhsUE0cjf+rIH6cpsZWsNziLAP4SaWEdIvIREMG/JjXthYrvuDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4c3c2ca30b4e75130d630d33ec0288b12cbca8e4c7a40e4326ca20cf7ebadad2","last_reissued_at":"2026-07-05T08:20:09.200403Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:20:09.200403Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2312.12321","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":"2026-07-05T08:20:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hfumdXHNdvlJ8T24CQIMdbvLJrtnTwDfHitrk17HAMdCA+vxNMSaFbwbfWYYyNCEdc4xm096ZW+amW6rPjpEBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T10:40:10.675634Z"},"content_sha256":"40449e0b631e9c9359dddbd1d1c267717f2ee5854229c2797144007de7be554b","schema_version":"1.0","event_id":"sha256:40449e0b631e9c9359dddbd1d1c267717f2ee5854229c2797144007de7be554b"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:JQ6CZIYLJZ2RGDLDBUZ6YAUIWE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Bypassing the Safety Training of Open-Source LLMs with Priming Attacks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CR","authors_text":"Changming Xu, Gagandeep Singh, Isha Chaudhary, Jason Vega","submitted_at":"2023-12-19T16:47:12Z","abstract_excerpt":"With the recent surge in popularity of LLMs has come an ever-increasing need for LLM safety training. In this paper, we investigate the fragility of SOTA open-source LLMs under simple, optimization-free attacks we refer to as $\\textit{priming attacks}$, which are easy to execute and effectively bypass alignment from safety training. Our proposed attack improves the Attack Success Rate on Harmful Behaviors, as measured by Llama Guard, by up to $3.3\\times$ compared to baselines. Source code and data are available at https://github.com/uiuc-focal-lab/llm-priming-attacks."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.12321","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/2312.12321/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":"2026-07-05T08:20:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"P4mUEyRx5cIf031+yqq5WWEWTkKU0i8Dp/Xy4TI08TBoyTIuauM+utqB5vuwTlu7djlJRbMBpFiU3M7qslgDDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T10:40:10.676016Z"},"content_sha256":"d33f9f298c34a5c18f0a10611e740561a705b41c1e3515c78a7d2b996a437dca","schema_version":"1.0","event_id":"sha256:d33f9f298c34a5c18f0a10611e740561a705b41c1e3515c78a7d2b996a437dca"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/JQ6CZIYLJZ2RGDLDBUZ6YAUIWE/bundle.json","state_url":"https://pith.science/pith/JQ6CZIYLJZ2RGDLDBUZ6YAUIWE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/JQ6CZIYLJZ2RGDLDBUZ6YAUIWE/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":"2026-08-04T10:40:10Z","links":{"resolver":"https://pith.science/pith/JQ6CZIYLJZ2RGDLDBUZ6YAUIWE","bundle":"https://pith.science/pith/JQ6CZIYLJZ2RGDLDBUZ6YAUIWE/bundle.json","state":"https://pith.science/pith/JQ6CZIYLJZ2RGDLDBUZ6YAUIWE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/JQ6CZIYLJZ2RGDLDBUZ6YAUIWE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:JQ6CZIYLJZ2RGDLDBUZ6YAUIWE","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":"68aeea68b057f65d05c3def9444b44116a1ee7b8ac93ab8da0bc77ca78af3c60","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2023-12-19T16:47:12Z","title_canon_sha256":"9adf5facd573caed217df7a48ed5fc17f5651353e17e6082d9151fbc09ad6e9c"},"schema_version":"1.0","source":{"id":"2312.12321","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.12321","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"arxiv_version","alias_value":"2312.12321v2","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.12321","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"pith_short_12","alias_value":"JQ6CZIYLJZ2R","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"pith_short_16","alias_value":"JQ6CZIYLJZ2RGDLD","created_at":"2026-07-05T08:20:09Z"},{"alias_kind":"pith_short_8","alias_value":"JQ6CZIYL","created_at":"2026-07-05T08:20:09Z"}],"graph_snapshots":[{"event_id":"sha256:d33f9f298c34a5c18f0a10611e740561a705b41c1e3515c78a7d2b996a437dca","target":"graph","created_at":"2026-07-05T08:20:09Z","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/2312.12321/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"With the recent surge in popularity of LLMs has come an ever-increasing need for LLM safety training. In this paper, we investigate the fragility of SOTA open-source LLMs under simple, optimization-free attacks we refer to as $\\textit{priming attacks}$, which are easy to execute and effectively bypass alignment from safety training. Our proposed attack improves the Attack Success Rate on Harmful Behaviors, as measured by Llama Guard, by up to $3.3\\times$ compared to baselines. Source code and data are available at https://github.com/uiuc-focal-lab/llm-priming-attacks.","authors_text":"Changming Xu, Gagandeep Singh, Isha Chaudhary, Jason Vega","cross_cats":["cs.AI","cs.CL","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2023-12-19T16:47:12Z","title":"Bypassing the Safety Training of Open-Source LLMs with Priming Attacks"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.12321","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":"sha256:40449e0b631e9c9359dddbd1d1c267717f2ee5854229c2797144007de7be554b","target":"record","created_at":"2026-07-05T08:20:09Z","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":"68aeea68b057f65d05c3def9444b44116a1ee7b8ac93ab8da0bc77ca78af3c60","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2023-12-19T16:47:12Z","title_canon_sha256":"9adf5facd573caed217df7a48ed5fc17f5651353e17e6082d9151fbc09ad6e9c"},"schema_version":"1.0","source":{"id":"2312.12321","kind":"arxiv","version":2}},"canonical_sha256":"4c3c2ca30b4e75130d630d33ec0288b12cbca8e4c7a40e4326ca20cf7ebadad2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4c3c2ca30b4e75130d630d33ec0288b12cbca8e4c7a40e4326ca20cf7ebadad2","first_computed_at":"2026-07-05T08:20:09.200403Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:20:09.200403Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"P8uuEqaIl3tqY8qah7aaJr4fbuTBFlsPCm/NhsUE0cjf+rIH6cpsZWsNziLAP4SaWEdIvIREMG/JjXthYrvuDQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:20:09.200835Z","signed_message":"canonical_sha256_bytes"},"source_id":"2312.12321","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:40449e0b631e9c9359dddbd1d1c267717f2ee5854229c2797144007de7be554b","sha256:d33f9f298c34a5c18f0a10611e740561a705b41c1e3515c78a7d2b996a437dca"],"state_sha256":"817d2191b1f049c789246016b121b271dfce0e8352160d67a6a0dfb5177fb8c7"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wRPHQUk9pIiQtv5RzTirNiKxckxzVy7buRF2y6BQ082uaikpMiHrtWFBw5UozpzYmYqYhqWk3P2N4cipOoQQDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T10:40:10.678814Z","bundle_sha256":"0d481d62f61d009c239837c8b85a29c6901b0bead001e38d9e13b63b876a3414"}}