mlsanityData to 7 Oct 2026

SAFER: Advancing Safety Alignment via Efficient Ex-Ante Reasoning

arXiv 2504.02725 ↗First posted 3 Apr 2025cs.CLcited by 4

On the map: LLM reasoning → Reasoning for agents, planning and code

also in 2 more
Abstract (arXiv)

Recent advancements in large language models (LLMs) have accelerated progress toward artificial general intelligence, yet their potential to generate harmful content poses critical safety challenges. Existing alignment methods often struggle to cover diverse safety scenarios and remain vulnerable to adversarial attacks. In this work, we propose SAFER, a framework for Safety Alignment via eFficient Ex-Ante Reasoning. Our approach instantiates structured Ex-Ante reasoning through initial assessment, rule verification, and path calibration, and embeds predefined safety rules to provide transparent and verifiable safety judgments. Specifically, our approach consists of two training stages: (1) supervised fine-tuning with synthetic traces to teach the multi-stage Ex-Ante reasoning, and (2) step-level reasoning preference optimization to jointly enhance safety, utility, and efficiency. Experiments on multiple open-source LLMs demonstrate that SAFER significantly enhances safety performance while maintaining helpfulness and response efficiency.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 52 of its 66 references that are in the map. First posted on arXiv 3 Apr 2025.

Builds on

52 of 66 references.

Closest ancestors · 4

Same direction, newest first

  1. 2025STAR-1: Safer Alignment of Reasoning LLMs with 1K Data2 Apr
  2. 2025STAIR: Improving Safety Alignment with Introspective Reasoning4 Feb
  3. 2024Deliberative Alignment: Reasoning Enables Safer Language Models20 Dec
  4. 2024How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMs12 Jan

Other references · 48

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  4. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  5. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  6. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  7. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  8. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  9. 2024Qwen2.5 Technical Reportcited by 4,713
  10. 2024GPT-4o System Cardcited by 4,467
  11. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  12. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
36 more
  1. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 2,914
  2. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,748
  3. 2024LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Modelscited by 1,442
  4. 2024HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusalcited by 1,330
  5. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!cited by 1,106
  6. 2023LIMA: Less Is More for Alignmentcited by 1,080
  7. 2024Gemma: Open Models Based on Gemini Research and Technologycited by 889
  8. 2023Gemini: A Family of Highly Capable Multimodal Modelscited by 810
  9. 2023AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Modelscited by 738
  10. 2024DeepSeek-V3 Technical Reportcited by 724
  11. 2020Aligning AI With Shared Human Valuescited by 706
  12. 2022Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learnedcited by 702
  13. 2023RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedbackcited by 615
  14. 2023XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Modelscited by 477
  15. 2023RRHF: Rank Responses to Align Language Models with Human Feedback without tearscited by 454
  16. 2024Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attackscited by 450
  17. 2024Safety Alignment Should Be Made More Than Just a Few Tokens Deepcited by 450
  18. 2023Catastrophic Jailbreak of Open-source LLMs via Exploiting Generationcited by 439
  19. 2024A StrongREJECT for Empty Jailbreakscited by 352
  20. 2024Measuring short-form factuality in large language modelscited by 345
  21. 2024PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preferencecited by 252
  22. 2023ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversationcited by 224
  23. 2023Removing RLHF Protections in GPT-4 via Fine-Tuningcited by 165
  24. 2023RAIN: Your Language Models Can Align Themselves without Finetuningcited by 149
  25. 2022Why Should Adversarial Perturbations be Imperceptible? Rethink the Research Paradigm in Adversarial NLPcited by 99
  26. 2022Language Generation Models Can Cause Harm: So What Can We Do About It? An Actionable Surveycited by 80
  27. 2023Managing extreme AI risks amid rapid progresscited by 48
  28. 2023Bypassing the Safety Training of Open-Source LLMs with Priming Attackscited by 48
  29. 2023BadLlama: cheaply removing safety fine-tuning from Llama 2-Chat 13Bcited by 35
  30. 2024Backtracking Improves Generation Safetycited by 31
  31. 2024LabSafety Bench: Benchmarking LLMs on Safety Issues in Scientific Labscited by 24
  32. 2024SciKnowEval: Evaluating Multi-level Scientific Knowledge of Large Language Modelscited by 19
  33. 2024SciSafeEval: A Comprehensive Benchmark for Safety Alignment of Large Language Models in Scientific Taskscited by 17
  34. 2024Reformatted Alignmentcited by 15
  35. 2024Course-Correction: Safety Alignment Using Synthetic Preferencescited by 13
  36. 2024Challenges in Guardrailing Large Language Models for Sciencecited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 3; 2026: 1.

4 papers.

  1. 2025A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Modelscited by 22
  2. 2025IntentionReasoner: Facilitating Adaptive LLM Safeguards through Intent Reasoning and Selective Query Refinementcited by 3
  3. 2025Strong but Brittle: Simple Attacks Subvert Reasoning-based Safety Guardrailscited by 2
  4. 2026Addressing Over-Refusal in LLMs with Competing Rewardscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025SaRO: Enhancing LLM Safety through Reasoning-based Alignment
  2. 2025RSafe: Incentivizing proactive reasoning to build robust and adaptive LLM safeguards
  3. 2025Safety Alignment Can Be Not Superficial With Explicit Safety Signals
  4. 2024Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements
  5. 2024Superficial Safety Alignment Hypothesis
  6. 2026Beyond Safe Data: Pretraining-Stage Alignment with Regular Safety Reflection
  7. 2026Reasoning over Precedents Alongside Statutes: Case-Augmented Deliberative Alignment for LLM Safety
  8. 2025Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing