{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:THMDJAIDA3PROWYF7NDSTZK6G6","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":"9fe552bedd5443865ca549fe797c085c2891793d22798310333fb041e02ac926","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-09-18T02:07:22Z","title_canon_sha256":"9d3be5c6edbceeed06179311464977d4c8b126875aa7b157ca0a274128ed53b1"},"schema_version":"1.0","source":{"id":"2309.14348","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2309.14348","created_at":"2026-07-05T08:30:29Z"},{"alias_kind":"arxiv_version","alias_value":"2309.14348v3","created_at":"2026-07-05T08:30:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2309.14348","created_at":"2026-07-05T08:30:29Z"},{"alias_kind":"pith_short_12","alias_value":"THMDJAIDA3PR","created_at":"2026-07-05T08:30:29Z"},{"alias_kind":"pith_short_16","alias_value":"THMDJAIDA3PROWYF","created_at":"2026-07-05T08:30:29Z"},{"alias_kind":"pith_short_8","alias_value":"THMDJAID","created_at":"2026-07-05T08:30:29Z"}],"graph_snapshots":[{"event_id":"sha256:0d68c73d32e05e55847523d3b45976bf078835cfbee98b574d4a5915e6c18929","target":"graph","created_at":"2026-07-05T08:30:29Z","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/2309.14348/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recently, Large Language Models (LLMs) have made significant advancements and are now widely used across various domains. Unfortunately, there has been a rising concern that LLMs can be misused to generate harmful or malicious content. Though a line of research has focused on aligning LLMs with human values and preventing them from producing inappropriate content, such alignments are usually vulnerable and can be bypassed by alignment-breaking attacks via adversarially optimized or handcrafted jailbreaking prompts. In this work, we introduce a Robustly Aligned LLM (RA-LLM) to defend against po","authors_text":"Bochuan Cao, Jinghui Chen, Lu Lin, Yuanpu Cao","cross_cats":["cs.AI","cs.CR","cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-09-18T02:07:22Z","title":"Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2309.14348","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":"sha256:4dd190ac9168e06d30e98b29243eea8c2fc22469d5de36e78bf2fcba4a58d691","target":"record","created_at":"2026-07-05T08:30:29Z","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":"9fe552bedd5443865ca549fe797c085c2891793d22798310333fb041e02ac926","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-09-18T02:07:22Z","title_canon_sha256":"9d3be5c6edbceeed06179311464977d4c8b126875aa7b157ca0a274128ed53b1"},"schema_version":"1.0","source":{"id":"2309.14348","kind":"arxiv","version":3}},"canonical_sha256":"99d834810306df175b05fb4729e55e37b6082cb5e2db760866d2c988d0889ca7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"99d834810306df175b05fb4729e55e37b6082cb5e2db760866d2c988d0889ca7","first_computed_at":"2026-07-05T08:30:29.766168Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:30:29.766168Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"lh0Wy6CQi4RnvfEWfsnIgKVc30EZc8FmMSlif+I4DqswcDPaVlEzHcaqZD4ck2MWRiidELhY+1lkNYx6y7b7AA==","signature_status":"signed_v1","signed_at":"2026-07-05T08:30:29.766931Z","signed_message":"canonical_sha256_bytes"},"source_id":"2309.14348","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4dd190ac9168e06d30e98b29243eea8c2fc22469d5de36e78bf2fcba4a58d691","sha256:0d68c73d32e05e55847523d3b45976bf078835cfbee98b574d4a5915e6c18929"],"state_sha256":"9e2c5282eeac960ccd5d5d2a07d633fff072a114e4a51327613064f6b29b583b"}