{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ZRC2VN4LYCRODTCGFLPXRX3KFV","short_pith_number":"pith:ZRC2VN4L","schema_version":"1.0","canonical_sha256":"cc45aab78bc0a2e1cc462adf78df6a2d746b48ce0a1b01c2c142db52453c0bc8","source":{"kind":"arxiv","id":"2409.11445","version":2},"attestation_state":"computed","paper":{"title":"Jailbreaking Large Language Models with Symbolic Mathematics","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CR","authors_text":"Emet Bethany, Juan Arturo Nolazco Flores, Mazal Bethany, Peyman Najafirad, Sumit Kumar Jha","submitted_at":"2024-09-17T03:39:45Z","abstract_excerpt":"Recent advancements in AI safety have led to increased efforts in training and red-teaming large language models (LLMs) to mitigate unsafe content generation. However, these safety mechanisms may not be comprehensive, leaving potential vulnerabilities unexplored. This paper introduces MathPrompt, a novel jailbreaking technique that exploits LLMs' advanced capabilities in symbolic mathematics to bypass their safety mechanisms. By encoding harmful natural language prompts into mathematical problems, we demonstrate a critical vulnerability in current AI safety measures. Our experiments across 13 "},"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":"2409.11445","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-09-17T03:39:45Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"8c42c2fe63d7d05d15df3a45c28e9b817f3b6f55e254c0b7b8173d2ca1ef9f14","abstract_canon_sha256":"bc78fef82314faed5e7140f0cc83e4e8c821dba1755c59ab596fc2e4f5ba1968"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:31:08.578712Z","signature_b64":"AFDGDPkVbkyLb634VuBVvX7Q7Q5rSGluq0pglaUO1QsaQpnFWuWHl4w0/JplJ7JiSuAVvPCg4PmnSjTP5vzBCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cc45aab78bc0a2e1cc462adf78df6a2d746b48ce0a1b01c2c142db52453c0bc8","last_reissued_at":"2026-07-05T09:31:08.578223Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:31:08.578223Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Jailbreaking Large Language Models with Symbolic Mathematics","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CR","authors_text":"Emet Bethany, Juan Arturo Nolazco Flores, Mazal Bethany, Peyman Najafirad, Sumit Kumar Jha","submitted_at":"2024-09-17T03:39:45Z","abstract_excerpt":"Recent advancements in AI safety have led to increased efforts in training and red-teaming large language models (LLMs) to mitigate unsafe content generation. However, these safety mechanisms may not be comprehensive, leaving potential vulnerabilities unexplored. This paper introduces MathPrompt, a novel jailbreaking technique that exploits LLMs' advanced capabilities in symbolic mathematics to bypass their safety mechanisms. By encoding harmful natural language prompts into mathematical problems, we demonstrate a critical vulnerability in current AI safety measures. Our experiments across 13 "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.11445","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/2409.11445/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":"2409.11445","created_at":"2026-07-05T09:31:08.578281+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.11445v2","created_at":"2026-07-05T09:31:08.578281+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.11445","created_at":"2026-07-05T09:31:08.578281+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZRC2VN4LYCRO","created_at":"2026-07-05T09:31:08.578281+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZRC2VN4LYCRODTCG","created_at":"2026-07-05T09:31:08.578281+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZRC2VN4L","created_at":"2026-07-05T09:31:08.578281+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.06921","citing_title":"Neuro-Symbolic AI for Cybersecurity: State of the Art, Challenges, and Opportunities","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03441","citing_title":"Exposing LLM Safety Gaps Through Mathematical Encoding:New Attacks and Systematic Analysis","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV","json":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV.json","graph_json":"https://pith.science/api/pith-number/ZRC2VN4LYCRODTCGFLPXRX3KFV/graph.json","events_json":"https://pith.science/api/pith-number/ZRC2VN4LYCRODTCGFLPXRX3KFV/events.json","paper":"https://pith.science/paper/ZRC2VN4L"},"agent_actions":{"view_html":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV","download_json":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV.json","view_paper":"https://pith.science/paper/ZRC2VN4L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.11445&json=true","fetch_graph":"https://pith.science/api/pith-number/ZRC2VN4LYCRODTCGFLPXRX3KFV/graph.json","fetch_events":"https://pith.science/api/pith-number/ZRC2VN4LYCRODTCGFLPXRX3KFV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV/action/storage_attestation","attest_author":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV/action/author_attestation","sign_citation":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV/action/citation_signature","submit_replication":"https://pith.science/pith/ZRC2VN4LYCRODTCGFLPXRX3KFV/action/replication_record"}},"created_at":"2026-07-05T09:31:08.578281+00:00","updated_at":"2026-07-05T09:31:08.578281+00:00"}