{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:6XNWX3T4I3RY7LS5LJYNO6VQKA","short_pith_number":"pith:6XNWX3T4","schema_version":"1.0","canonical_sha256":"f5db6bee7c46e38fae5d5a70d77ab0502e5056c965c0d7f24746b7c2c9e53753","source":{"kind":"arxiv","id":"2402.09497","version":2},"attestation_state":"computed","paper":{"title":"Instruction Tuning for Secure Code Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","cs.SE"],"primary_cat":"cs.CR","authors_text":"Gabriela Krasnopolska, Jingxuan He, Mark Vero, Martin Vechev","submitted_at":"2024-02-14T15:47:46Z","abstract_excerpt":"Modern language models (LMs) have gained widespread acceptance in everyday and professional contexts, particularly in programming. An essential procedure enabling this adoption is instruction tuning, which substantially enhances LMs' practical utility by training them to follow user instructions and human preferences. However, existing instruction tuning schemes overlook a crucial aspect: the security of generated code. As a result, even the state-of-the-art instruction-tuned LMs frequently produce unsafe code, posing significant security risks. In this work, we introduce SafeCoder to address "},"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":"2402.09497","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-02-14T15:47:46Z","cross_cats_sorted":["cs.AI","cs.LG","cs.SE"],"title_canon_sha256":"fa88d66f8f8f301818d0c19c368f2eb75cf46fa92e06f2276280ad534157ca43","abstract_canon_sha256":"18872968038ad6f73b48edf02c994a51916188efae0d43cb08676200d3faf0f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:43:08.530024Z","signature_b64":"nzQBUxjartH9O+YJ42Krdu2v1ixxHXdqaLpimtiCuBGe4aVGDX/9kEgsUjGSaeL9YIntjh/6jAl3wmeY+R0wCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f5db6bee7c46e38fae5d5a70d77ab0502e5056c965c0d7f24746b7c2c9e53753","last_reissued_at":"2026-07-05T08:43:08.529595Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:43:08.529595Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Instruction Tuning for Secure Code Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","cs.SE"],"primary_cat":"cs.CR","authors_text":"Gabriela Krasnopolska, Jingxuan He, Mark Vero, Martin Vechev","submitted_at":"2024-02-14T15:47:46Z","abstract_excerpt":"Modern language models (LMs) have gained widespread acceptance in everyday and professional contexts, particularly in programming. An essential procedure enabling this adoption is instruction tuning, which substantially enhances LMs' practical utility by training them to follow user instructions and human preferences. However, existing instruction tuning schemes overlook a crucial aspect: the security of generated code. As a result, even the state-of-the-art instruction-tuned LMs frequently produce unsafe code, posing significant security risks. In this work, we introduce SafeCoder to address "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.09497","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/2402.09497/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":"2402.09497","created_at":"2026-07-05T08:43:08.529655+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.09497v2","created_at":"2026-07-05T08:43:08.529655+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.09497","created_at":"2026-07-05T08:43:08.529655+00:00"},{"alias_kind":"pith_short_12","alias_value":"6XNWX3T4I3RY","created_at":"2026-07-05T08:43:08.529655+00:00"},{"alias_kind":"pith_short_16","alias_value":"6XNWX3T4I3RY7LS5","created_at":"2026-07-05T08:43:08.529655+00:00"},{"alias_kind":"pith_short_8","alias_value":"6XNWX3T4","created_at":"2026-07-05T08:43:08.529655+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25973","citing_title":"Helpful or Harmful? Evaluating LLM-Assisted Vulnerability Patching via a Human Study","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11755","citing_title":"Acoda: Adversarial Code Obfuscation for Defending against LLM-based Analysis","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29239","citing_title":"Breaking the Rounding Trap: Securing LLMs against Quantization-Conditioned Backdoors","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08382","citing_title":"SecureForge: Finding and Preventing Vulnerabilities in LLM-Generated Code via Prompt Optimization","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12214","citing_title":"Structural Anchors and Reasoning Fragility:Understanding CoT Robustness in LLM4Code","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03179","citing_title":"A Validated Prompt Bank for Malicious Code Generation: Separating Executable Weapons from Security Knowledge in 1,554 Consensus-Labeled Prompts","ref_index":51,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA","json":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA.json","graph_json":"https://pith.science/api/pith-number/6XNWX3T4I3RY7LS5LJYNO6VQKA/graph.json","events_json":"https://pith.science/api/pith-number/6XNWX3T4I3RY7LS5LJYNO6VQKA/events.json","paper":"https://pith.science/paper/6XNWX3T4"},"agent_actions":{"view_html":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA","download_json":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA.json","view_paper":"https://pith.science/paper/6XNWX3T4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.09497&json=true","fetch_graph":"https://pith.science/api/pith-number/6XNWX3T4I3RY7LS5LJYNO6VQKA/graph.json","fetch_events":"https://pith.science/api/pith-number/6XNWX3T4I3RY7LS5LJYNO6VQKA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA/action/storage_attestation","attest_author":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA/action/author_attestation","sign_citation":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA/action/citation_signature","submit_replication":"https://pith.science/pith/6XNWX3T4I3RY7LS5LJYNO6VQKA/action/replication_record"}},"created_at":"2026-07-05T08:43:08.529655+00:00","updated_at":"2026-07-05T08:43:08.529655+00:00"}