{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:DDOZ3WKPSOSD2JDLG72XGMXZLM","short_pith_number":"pith:DDOZ3WKP","canonical_record":{"source":{"id":"2310.06387","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-10T07:50:29Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CR"],"title_canon_sha256":"31f7bb9afcc76557695448e50fc0eccfce6ad703e4fbe6df55a883896e185eac","abstract_canon_sha256":"15dfd35c1215754c449dd13b3e56bd284060ef3239b8d06538899af3653532e4"},"schema_version":"1.0"},"canonical_sha256":"18dd9dd94f93a43d246b37f57332f95b0790c17fde9b0b11de7ae843e8915bf2","source":{"kind":"arxiv","id":"2310.06387","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.06387","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"arxiv_version","alias_value":"2310.06387v3","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.06387","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"pith_short_12","alias_value":"DDOZ3WKPSOSD","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"pith_short_16","alias_value":"DDOZ3WKPSOSD2JDL","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"pith_short_8","alias_value":"DDOZ3WKP","created_at":"2026-07-05T08:23:12Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:DDOZ3WKPSOSD2JDLG72XGMXZLM","target":"record","payload":{"canonical_record":{"source":{"id":"2310.06387","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-10T07:50:29Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CR"],"title_canon_sha256":"31f7bb9afcc76557695448e50fc0eccfce6ad703e4fbe6df55a883896e185eac","abstract_canon_sha256":"15dfd35c1215754c449dd13b3e56bd284060ef3239b8d06538899af3653532e4"},"schema_version":"1.0"},"canonical_sha256":"18dd9dd94f93a43d246b37f57332f95b0790c17fde9b0b11de7ae843e8915bf2","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:23:12.845400Z","signature_b64":"iC8zzuO1OnKiyvnmC5fB8C6vMktkk12yScFOIsIhP4fXXJ2vHdmFoqR+5d+smEaDfIdFQmBjYApJs6kkaUwxBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"18dd9dd94f93a43d246b37f57332f95b0790c17fde9b0b11de7ae843e8915bf2","last_reissued_at":"2026-07-05T08:23:12.844865Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:23:12.844865Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2310.06387","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:23:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PbNIejuBwk/qDGN4jHUDm3P56Cp1hOLLKiSduuK32qDqrxFC55zSgh7pusktF0eaYksyxMEVRM+pUpwBYmOcBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-22T01:08:59.589038Z"},"content_sha256":"f88030b0d9c48eb42cc2d21770d03c9e3722d142e0ef43899f1086e38402302b","schema_version":"1.0","event_id":"sha256:f88030b0d9c48eb42cc2d21770d03c9e3722d142e0ef43899f1086e38402302b"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:DDOZ3WKPSOSD2JDLG72XGMXZLM","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Ang Li, Yichuan Mo, Yifei Wang, Yisen Wang, Zeming Wei","submitted_at":"2023-10-10T07:50:29Z","abstract_excerpt":"Large Language Models (LLMs) have shown remarkable success in various tasks, yet their safety and the risk of generating harmful content remain pressing concerns. In this paper, we delve into the potential of In-Context Learning (ICL) to modulate the alignment of LLMs. Specifically, we propose the In-Context Attack (ICA) which employs harmful demonstrations to subvert LLMs, and the In-Context Defense (ICD) which bolsters model resilience through examples that demonstrate refusal to produce harmful responses. We offer theoretical insights to elucidate how a limited set of in-context demonstrati"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.06387","kind":"arxiv","version":3},"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/2310.06387/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":"2026-07-05T08:23:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RZ52Jetcrk8GJH9dRe3fp/lzIhgr9mCuBOZ56DzIM/cA0kOw2bYN8idk/W7mXjkRc9cw7+/CJC4ResamRvrvDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-22T01:08:59.589425Z"},"content_sha256":"9c52d4a3e7ca8c365a08b25e452fbb4b0a762c39dde7cb18922922b87a4bd410","schema_version":"1.0","event_id":"sha256:9c52d4a3e7ca8c365a08b25e452fbb4b0a762c39dde7cb18922922b87a4bd410"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DDOZ3WKPSOSD2JDLG72XGMXZLM/bundle.json","state_url":"https://pith.science/pith/DDOZ3WKPSOSD2JDLG72XGMXZLM/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DDOZ3WKPSOSD2JDLG72XGMXZLM/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":"2026-07-22T01:08:59Z","links":{"resolver":"https://pith.science/pith/DDOZ3WKPSOSD2JDLG72XGMXZLM","bundle":"https://pith.science/pith/DDOZ3WKPSOSD2JDLG72XGMXZLM/bundle.json","state":"https://pith.science/pith/DDOZ3WKPSOSD2JDLG72XGMXZLM/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DDOZ3WKPSOSD2JDLG72XGMXZLM/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:DDOZ3WKPSOSD2JDLG72XGMXZLM","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":"15dfd35c1215754c449dd13b3e56bd284060ef3239b8d06538899af3653532e4","cross_cats_sorted":["cs.AI","cs.CL","cs.CR"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-10T07:50:29Z","title_canon_sha256":"31f7bb9afcc76557695448e50fc0eccfce6ad703e4fbe6df55a883896e185eac"},"schema_version":"1.0","source":{"id":"2310.06387","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.06387","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"arxiv_version","alias_value":"2310.06387v3","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.06387","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"pith_short_12","alias_value":"DDOZ3WKPSOSD","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"pith_short_16","alias_value":"DDOZ3WKPSOSD2JDL","created_at":"2026-07-05T08:23:12Z"},{"alias_kind":"pith_short_8","alias_value":"DDOZ3WKP","created_at":"2026-07-05T08:23:12Z"}],"graph_snapshots":[{"event_id":"sha256:9c52d4a3e7ca8c365a08b25e452fbb4b0a762c39dde7cb18922922b87a4bd410","target":"graph","created_at":"2026-07-05T08:23:12Z","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/2310.06387/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) have shown remarkable success in various tasks, yet their safety and the risk of generating harmful content remain pressing concerns. In this paper, we delve into the potential of In-Context Learning (ICL) to modulate the alignment of LLMs. Specifically, we propose the In-Context Attack (ICA) which employs harmful demonstrations to subvert LLMs, and the In-Context Defense (ICD) which bolsters model resilience through examples that demonstrate refusal to produce harmful responses. We offer theoretical insights to elucidate how a limited set of in-context demonstrati","authors_text":"Ang Li, Yichuan Mo, Yifei Wang, Yisen Wang, Zeming Wei","cross_cats":["cs.AI","cs.CL","cs.CR"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-10T07:50:29Z","title":"Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.06387","kind":"arxiv","version":3},"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:f88030b0d9c48eb42cc2d21770d03c9e3722d142e0ef43899f1086e38402302b","target":"record","created_at":"2026-07-05T08:23:12Z","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":"15dfd35c1215754c449dd13b3e56bd284060ef3239b8d06538899af3653532e4","cross_cats_sorted":["cs.AI","cs.CL","cs.CR"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-10T07:50:29Z","title_canon_sha256":"31f7bb9afcc76557695448e50fc0eccfce6ad703e4fbe6df55a883896e185eac"},"schema_version":"1.0","source":{"id":"2310.06387","kind":"arxiv","version":3}},"canonical_sha256":"18dd9dd94f93a43d246b37f57332f95b0790c17fde9b0b11de7ae843e8915bf2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"18dd9dd94f93a43d246b37f57332f95b0790c17fde9b0b11de7ae843e8915bf2","first_computed_at":"2026-07-05T08:23:12.844865Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:23:12.844865Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"iC8zzuO1OnKiyvnmC5fB8C6vMktkk12yScFOIsIhP4fXXJ2vHdmFoqR+5d+smEaDfIdFQmBjYApJs6kkaUwxBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:23:12.845400Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.06387","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f88030b0d9c48eb42cc2d21770d03c9e3722d142e0ef43899f1086e38402302b","sha256:9c52d4a3e7ca8c365a08b25e452fbb4b0a762c39dde7cb18922922b87a4bd410"],"state_sha256":"444b5b6aca9ad17c2198c77d5165f321b97d45e6b61c408b19c8c25cffd84b5c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FVUL7d5MGQZL4Y7VHE5NkwyjvrxpbpEEU9YZWevBxsxYb8z6qw1nu6eaOV1SBg0jW2zvPvNCFD44xmlWdsxRCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-22T01:08:59.591566Z","bundle_sha256":"d88562a5c896251de73e1545cb5215a8d774263c2c2dddffa66d395803de0674"}}