{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:US2QW4MRWXVFN2WGY4F6SH6AXD","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":"70858113067127889c05db46d5e3db69d7de2205ecd17ae821d55bdfbb81012f","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-12-08T01:41:36Z","title_canon_sha256":"3a2aab6df883ad70eb28e0a7c1f6fb9d8a8b860a1a9a01329794eefe8eaf46cb"},"schema_version":"1.0","source":{"id":"2312.04782","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.04782","created_at":"2026-07-05T07:21:51Z"},{"alias_kind":"arxiv_version","alias_value":"2312.04782v1","created_at":"2026-07-05T07:21:51Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.04782","created_at":"2026-07-05T07:21:51Z"},{"alias_kind":"pith_short_12","alias_value":"US2QW4MRWXVF","created_at":"2026-07-05T07:21:51Z"},{"alias_kind":"pith_short_16","alias_value":"US2QW4MRWXVFN2WG","created_at":"2026-07-05T07:21:51Z"},{"alias_kind":"pith_short_8","alias_value":"US2QW4MR","created_at":"2026-07-05T07:21:51Z"}],"graph_snapshots":[{"event_id":"sha256:ed716bb97a4f41a5e7aba26cb9fae6ce570175f6ee4b6b381ca3012f8af4fe09","target":"graph","created_at":"2026-07-05T07:21:51Z","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/2312.04782/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) are now widely used in various applications, making it crucial to align their ethical standards with human values. However, recent jail-breaking methods demonstrate that this alignment can be undermined using carefully constructed prompts. In our study, we reveal a new threat to LLM alignment when a bad actor has access to the model's output logits, a common feature in both open-source LLMs and many commercial LLM APIs (e.g., certain GPT models). It does not rely on crafting specific prompts. Instead, it exploits the fact that even when an LLM rejects a toxic reque","authors_text":"Guangyu Shen, Guanhong Tao, Siyuan Cheng, Xiangyu Zhang, Zhuo Zhang","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-12-08T01:41:36Z","title":"Make Them Spill the Beans! Coercive Knowledge Extraction from (Production) LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.04782","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":"sha256:735434a6488ef5f2fb45567c01be0175e38f90e706711da92f2b473de54f8a91","target":"record","created_at":"2026-07-05T07:21:51Z","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":"70858113067127889c05db46d5e3db69d7de2205ecd17ae821d55bdfbb81012f","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-12-08T01:41:36Z","title_canon_sha256":"3a2aab6df883ad70eb28e0a7c1f6fb9d8a8b860a1a9a01329794eefe8eaf46cb"},"schema_version":"1.0","source":{"id":"2312.04782","kind":"arxiv","version":1}},"canonical_sha256":"a4b50b7191b5ea56eac6c70be91fc0b8e53990b4ed5fce42a749543ad0cdb24b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a4b50b7191b5ea56eac6c70be91fc0b8e53990b4ed5fce42a749543ad0cdb24b","first_computed_at":"2026-07-05T07:21:51.345670Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:21:51.345670Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"D+trCFVDP3pA7KkRcLqWNyZQqYbPDvujfVTnxlIT5boQpWA50S2IGErQvOaZ0Xyj6x0jRyoO7vFgQCdMA8+QBw==","signature_status":"signed_v1","signed_at":"2026-07-05T07:21:51.346910Z","signed_message":"canonical_sha256_bytes"},"source_id":"2312.04782","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:735434a6488ef5f2fb45567c01be0175e38f90e706711da92f2b473de54f8a91","sha256:ed716bb97a4f41a5e7aba26cb9fae6ce570175f6ee4b6b381ca3012f8af4fe09"],"state_sha256":"bd9f4fd67a976b1cdd2b2708618b909828b6f57901439ab60feceaeff19426df"}