{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:REO4SNA55WJ6ORFVAUKQMNJ3IG","short_pith_number":"pith:REO4SNA5","schema_version":"1.0","canonical_sha256":"891dc9341ded93e744b5051506353b41a9a6f98e94e2c09ace088a19f086ffa4","source":{"kind":"arxiv","id":"2411.09125","version":1},"attestation_state":"computed","paper":{"title":"DROJ: A Prompt-Driven Attack against Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"BoRan Wang, Leyang Hu","submitted_at":"2024-11-14T01:48:08Z","abstract_excerpt":"Large Language Models (LLMs) have demonstrated exceptional capabilities across various natural language processing tasks. Due to their training on internet-sourced datasets, LLMs can sometimes generate objectionable content, necessitating extensive alignment with human feedback to avoid such outputs. Despite massive alignment efforts, LLMs remain susceptible to adversarial jailbreak attacks, which usually are manipulated prompts designed to circumvent safety mechanisms and elicit harmful responses. Here, we introduce a novel approach, Directed Rrepresentation Optimization Jailbreak (DROJ), whi"},"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":"2411.09125","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-11-14T01:48:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"86ef766198a5a1fbad432d64ab8153e570547a1773a869cfb4649912c782e500","abstract_canon_sha256":"1af5127679c602b46b415cb7ee70f30048966fb2ab692a328cad234a7ff9fca7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:35:20.559125Z","signature_b64":"XASe/50mEFWsR3Lha+bmoIEAuUffERpLWMhnnE4B4X/I5AetaE/NuhVDcFYjnwEStegUiNNMia6a9xy1JqEAAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"891dc9341ded93e744b5051506353b41a9a6f98e94e2c09ace088a19f086ffa4","last_reissued_at":"2026-07-05T09:35:20.558335Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:35:20.558335Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DROJ: A Prompt-Driven Attack against Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"BoRan Wang, Leyang Hu","submitted_at":"2024-11-14T01:48:08Z","abstract_excerpt":"Large Language Models (LLMs) have demonstrated exceptional capabilities across various natural language processing tasks. Due to their training on internet-sourced datasets, LLMs can sometimes generate objectionable content, necessitating extensive alignment with human feedback to avoid such outputs. Despite massive alignment efforts, LLMs remain susceptible to adversarial jailbreak attacks, which usually are manipulated prompts designed to circumvent safety mechanisms and elicit harmful responses. Here, we introduce a novel approach, Directed Rrepresentation Optimization Jailbreak (DROJ), whi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.09125","kind":"arxiv","version":1},"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/2411.09125/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":"2411.09125","created_at":"2026-07-05T09:35:20.558399+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.09125v1","created_at":"2026-07-05T09:35:20.558399+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.09125","created_at":"2026-07-05T09:35:20.558399+00:00"},{"alias_kind":"pith_short_12","alias_value":"REO4SNA55WJ6","created_at":"2026-07-05T09:35:20.558399+00:00"},{"alias_kind":"pith_short_16","alias_value":"REO4SNA55WJ6ORFV","created_at":"2026-07-05T09:35:20.558399+00:00"},{"alias_kind":"pith_short_8","alias_value":"REO4SNA5","created_at":"2026-07-05T09:35:20.558399+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07539","citing_title":"Prompt Governance? On Governing Technologies Governed by Natural Language","ref_index":137,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG","json":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG.json","graph_json":"https://pith.science/api/pith-number/REO4SNA55WJ6ORFVAUKQMNJ3IG/graph.json","events_json":"https://pith.science/api/pith-number/REO4SNA55WJ6ORFVAUKQMNJ3IG/events.json","paper":"https://pith.science/paper/REO4SNA5"},"agent_actions":{"view_html":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG","download_json":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG.json","view_paper":"https://pith.science/paper/REO4SNA5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.09125&json=true","fetch_graph":"https://pith.science/api/pith-number/REO4SNA55WJ6ORFVAUKQMNJ3IG/graph.json","fetch_events":"https://pith.science/api/pith-number/REO4SNA55WJ6ORFVAUKQMNJ3IG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG/action/storage_attestation","attest_author":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG/action/author_attestation","sign_citation":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG/action/citation_signature","submit_replication":"https://pith.science/pith/REO4SNA55WJ6ORFVAUKQMNJ3IG/action/replication_record"}},"created_at":"2026-07-05T09:35:20.558399+00:00","updated_at":"2026-07-05T09:35:20.558399+00:00"}