{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/stair-improving-safety-alignment-with","title":"STAIR: Improving Safety Alignment with Introspective Reasoning","arxiv_id":"2502.02384","date":"2025-02-04","proceeding":null,"authors":["Yichi Zhang","Siyuan Zhang","Yao Huang","Zeyu Xia","Zhengwei Fang","Xiao Yang","Ranjie Duan","Dong Yan","Yinpeng Dong","Jun Zhu"],"abstract":"Ensuring the safety and harmlessness of Large Language Models (LLMs) has become equally critical as their performance in applications. However, existing safety alignment methods typically suffer from safety-performance trade-offs and the susceptibility to jailbreak attacks, primarily due to their reliance on direct refusals for malicious queries. In this paper, we propose STAIR, a novel framework that integrates SafeTy Alignment with Itrospective Reasoning. We enable LLMs to identify safety risks through step-by-step analysis by self-improving chain-of-thought (CoT) reasoning with safety awareness. STAIR first equips the model with a structured reasoning capability and then advances safety alignment via iterative preference optimization on step-level reasoning data generated using our newly proposed Safety-Informed Monte Carlo Tree Search (SI-MCTS). We further train a process reward model on this data to guide test-time searches for improved responses. Extensive experiments show that STAIR effectively mitigates harmful outputs while better preserving helpfulness, compared to instinctive alignment strategies. With test-time scaling, STAIR achieves a safety performance comparable to Claude-3.5 against popular jailbreak attacks. Relevant resources in this work are available at https://github.com/thu-ml/STAIR.","url_abs":"https://arxiv.org/abs/2502.02384v1","url_pdf":"https://arxiv.org/pdf/2502.02384v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"stair-improving-safety-alignment-with","repo_url":"https://github.com/thu-ml/stair","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"safety-alignment","task_name":"Safety Alignment"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2502.02384","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2502.02384"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/thu-ml/stair","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran_draft_wrong":1,"unverified":5},"by_repo_kind":{"official":{"samples":6,"ran":1,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"f8c65c92ec213846","entry":"read_json","repo":"thu-ml/stair","repo_kind":"official","path":"src/utils.py","file_url":"https://github.com/thu-ml/stair/blob/HEAD/src/utils.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f8c65c92ec213846"}},{"code_sha256_prefix":"0396dbaaf06aef8f","entry":"apply_template","repo":"thu-ml/stair","repo_kind":"official","path":"src/final_orm.py","file_url":"https://github.com/thu-ml/stair/blob/HEAD/src/final_orm.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0396dbaaf06aef8f"}},{"code_sha256_prefix":"fcfc0ecd7309f71a","entry":"extract_answer","repo":"thu-ml/stair","repo_kind":"official","path":"src/final_orm.py","file_url":"https://github.com/thu-ml/stair/blob/HEAD/src/final_orm.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"fcfc0ecd7309f71a"}},{"code_sha256_prefix":"1027b0caeb3fa262","entry":"read_json_lines","repo":"thu-ml/stair","repo_kind":"official","path":"src/utils.py","file_url":"https://github.com/thu-ml/stair/blob/HEAD/src/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1027b0caeb3fa262"}},{"code_sha256_prefix":"da32887c2ebcea3a","entry":"rule_generate","repo":"thu-ml/stair","repo_kind":"official","path":"src/final_orm.py","file_url":"https://github.com/thu-ml/stair/blob/HEAD/src/final_orm.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"da32887c2ebcea3a"}},{"code_sha256_prefix":"e270b1c7818dd070","entry":"trajectory_to_response","repo":"thu-ml/stair","repo_kind":"official","path":"src/utils.py","file_url":"https://github.com/thu-ml/stair/blob/HEAD/src/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"e270b1c7818dd070"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}