{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:WFFYBSCXZN5ZTLDEMVWVMXN7SZ","short_pith_number":"pith:WFFYBSCX","schema_version":"1.0","canonical_sha256":"b14b80c857cb7b99ac64656d565dbf9649a2c6cf6a7d569466154a40bea04195","source":{"kind":"arxiv","id":"2405.07932","version":2},"attestation_state":"computed","paper":{"title":"PARDEN, Can You Repeat That? Defending against Jailbreaks via Repetition","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jakob Foerster, Qizhen Zhang, Ziyang Zhang","submitted_at":"2024-05-13T17:08:42Z","abstract_excerpt":"Large language models (LLMs) have shown success in many natural language processing tasks. Despite rigorous safety alignment processes, supposedly safety-aligned LLMs like Llama 2 and Claude 2 are still susceptible to jailbreaks, leading to security risks and abuse of the models. One option to mitigate such risks is to augment the LLM with a dedicated \"safeguard\", which checks the LLM's inputs or outputs for undesired behaviour. A promising approach is to use the LLM itself as the safeguard. Nonetheless, baseline methods, such as prompting the LLM to self-classify toxic content, demonstrate li"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2405.07932","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-05-13T17:08:42Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b73851f0b8ee0e641479ab88988978766f3dfb299239a850142096b20a0cbe1e","abstract_canon_sha256":"9543f5d62d6d538520206fe3ed4bb43b6008cc73ccff29139eb55759a99d3400"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:18:53.651197Z","signature_b64":"1p9V0iVunCXY//mVX0dDH+vt6YP4ghjC8m60JtWkiVTOJAcofst/t3Z3J7VKvr+W0K5VaYqlV/CXp5G2q3b7DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b14b80c857cb7b99ac64656d565dbf9649a2c6cf6a7d569466154a40bea04195","last_reissued_at":"2026-07-05T08:18:53.650682Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:18:53.650682Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PARDEN, Can You Repeat That? Defending against Jailbreaks via Repetition","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jakob Foerster, Qizhen Zhang, Ziyang Zhang","submitted_at":"2024-05-13T17:08:42Z","abstract_excerpt":"Large language models (LLMs) have shown success in many natural language processing tasks. Despite rigorous safety alignment processes, supposedly safety-aligned LLMs like Llama 2 and Claude 2 are still susceptible to jailbreaks, leading to security risks and abuse of the models. One option to mitigate such risks is to augment the LLM with a dedicated \"safeguard\", which checks the LLM's inputs or outputs for undesired behaviour. A promising approach is to use the LLM itself as the safeguard. Nonetheless, baseline methods, such as prompting the LLM to self-classify toxic content, demonstrate li"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.07932","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/2405.07932/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2405.07932","created_at":"2026-07-05T08:18:53.650739+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.07932v2","created_at":"2026-07-05T08:18:53.650739+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.07932","created_at":"2026-07-05T08:18:53.650739+00:00"},{"alias_kind":"pith_short_12","alias_value":"WFFYBSCXZN5Z","created_at":"2026-07-05T08:18:53.650739+00:00"},{"alias_kind":"pith_short_16","alias_value":"WFFYBSCXZN5ZTLDE","created_at":"2026-07-05T08:18:53.650739+00:00"},{"alias_kind":"pith_short_8","alias_value":"WFFYBSCX","created_at":"2026-07-05T08:18:53.650739+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2502.05206","citing_title":"Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety","ref_index":116,"is_internal_anchor":false},{"citing_arxiv_id":"2602.02280","citing_title":"RACC: Representation-Aware Coverage Criteria for LLM Safety Testing","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04060","citing_title":"CoopGuard: Stateful Cooperative Agents Safeguarding LLMs Against Evolving Multi-Round Attacks","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24082","citing_title":"Jailbreaking Frontier Foundation Models Through Intention Deception","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ","json":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ.json","graph_json":"https://pith.science/api/pith-number/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/graph.json","events_json":"https://pith.science/api/pith-number/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/events.json","paper":"https://pith.science/paper/WFFYBSCX"},"agent_actions":{"view_html":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ","download_json":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ.json","view_paper":"https://pith.science/paper/WFFYBSCX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.07932&json=true","fetch_graph":"https://pith.science/api/pith-number/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/graph.json","fetch_events":"https://pith.science/api/pith-number/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/action/storage_attestation","attest_author":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/action/author_attestation","sign_citation":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/action/citation_signature","submit_replication":"https://pith.science/pith/WFFYBSCXZN5ZTLDEMVWVMXN7SZ/action/replication_record"}},"created_at":"2026-07-05T08:18:53.650739+00:00","updated_at":"2026-07-05T08:18:53.650739+00:00"}