mlsanityData to 7 Oct 2026

RealSafe-R1: Safety-Aligned DeepSeek-R1 without Compromising Reasoning Capability

arXiv 2504.10081 ↗First posted 14 Apr 2025cs.AI cs.CLcited by 58

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

also in 1 more
Abstract (arXiv)

Large Reasoning Models (LRMs), such as OpenAI o1 and DeepSeek-R1, have been rapidly progressing and achieving breakthrough performance on complex reasoning tasks such as mathematics and coding. However, the open-source R1 models have raised safety concerns in wide applications, such as the tendency to comply with malicious queries, which greatly impacts the utility of these powerful models in their applications. In this paper, we introduce RealSafe-R1 as safety-aligned versions of DeepSeek-R1 distilled models. To train these models, we construct a dataset of 15k safety-aware reasoning trajectories generated by DeepSeek-R1, under explicit instructions for expected refusal behavior. Both quantitative experiments and qualitative case studies demonstrate the models' improvements, which are shown in their safety guardrails against both harmful queries and jailbreak attacks. Importantly, unlike prior safety alignment efforts that often compromise reasoning performance, our method preserves the models' reasoning capabilities by maintaining the training data within the original distribution of generation. Model weights of RealSafe-R1 are open-source at https://huggingface.co/RealSafe.

Key facts. Cited by 58 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 27 of its 47 references that are in the map. First posted on arXiv 14 Apr 2025.

Builds on

27 of 47 references.

Closest ancestors · 10

Same direction, newest first

  1. 2025STAR-1: Safer Alignment of Reasoning LLMs with 1K Data2 Apr
  2. 2025Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable1 Mar
  3. 2025The Hidden Risks of Large Reasoning Models: A Safety Assessment of R118 Feb
  4. 2025SafeChain: Safety of Language Models with Long Chain-of-Thought Reasoning Capabilities17 Feb
  5. 2025Safety Evaluation of DeepSeek Models in Chinese Contexts16 Feb
  6. 2025STAIR: Improving Safety Alignment with Introspective Reasoning4 Feb
  7. 2024OpenAI o1 System Card21 Dec
  8. 2024Deliberative Alignment: Reasoning Enables Safer Language Models20 Dec
  9. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code12 Mar
  10. 2024CodeMind: Evaluating Large Language Models for Code Reasoning15 Feb

Other references · 17

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  4. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  5. 2023Let's Verify Step by Stepcited by 3,598
  6. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,919
  7. 2023Jailbreaking Black Box Large Language Models in Twenty Queriescited by 1,326
  8. 2023XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Modelscited by 477
  9. 2015Policy Distillationcited by 414
  10. 2023What Makes Good Data for Alignment? A Comprehensive Study of Automatic Data Selection in Instruction Tuningcited by 354
  11. 2024A StrongREJECT for Empty Jailbreakscited by 352
  12. 2024PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preferencecited by 252
5 more
  1. 2024JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attackscited by 208
  2. 2024AlphaMath Almost Zero: Process Supervision without Processcited by 160
  3. 2024Simulating Classroom Education with LLM-Empowered Agentscited by 125
  4. 2024Step-level Value Preference Optimization for Mathematical Reasoningcited by 71
  5. 2024Enhancing LLM Safety via Constrained Direct Preference Optimizationcited by 47

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 39; 2026: 19.

58 papers.

Sort:

By area:

  1. 2025The Landscape of Agentic Reinforcement Learning for LLMs: A Surveycited by 195
  2. 2025Safety in Large Reasoning Models: A Surveycited by 58
  3. 2025Mitigating Overthinking in Large Reasoning Models via Manifold Steeringcited by 32
  4. 2025A Survey on Autonomy-Induced Security Risks in Large Model-Based Agentscited by 28
  5. 2025SAFEPATH: Preventing Harmful Reasoning in Chain-of-Thought via Early Alignmentcited by 24
  6. 2025Reasoning as an Adaptive Defense for Safetycited by 22
  7. 2025A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Modelscited by 22
  8. 2025SafeKey: Amplifying Aha-Moment Insights for Safety Reasoningcited by 20
  9. 2025Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Modelscited by 19
  10. 2025A Survey of Slow Thinking-based Reasoning LLMs using Reinforced Learning and Inference-time Scaling Lawcited by 18
  11. 2025How Should We Enhance the Safety of Large Reasoning Models: An Empirical Studycited by 18
  12. 2025Breaking the Ceiling: Exploring the Potential of Jailbreak Attacks through Expanding Strategy Spacecited by 15
38 more
  1. 2025DeceptionBench: A Comprehensive Benchmark for AI Deception Behaviors in Real-world Scenarioscited by 14
  2. 2025Large Reasoning Models Learn Better Alignment from Flawed Thinkingcited by 13
  3. 2025Towards Safe Reasoning in Large Reasoning Models via Corrective Interventioncited by 11
  4. 2025Refusal Falls off a Cliff: How Safety Alignment Fails in Reasoning?cited by 11
  5. 2025SafeGRPO: Self-Rewarded Multimodal Safety Alignment via Rule-Governed Policy Optimizationcited by 11
  6. 2025ARMOR: Aligning Secure and Safe Large Language Models via Meticulous Reasoningcited by 8
  7. 2025Beyond Safe Answers: A Benchmark for Evaluating True Risk Awareness in Large Reasoning Modelscited by 7
  8. 2025UnsafeChain: Enhancing Reasoning Model Safety via Hard Casescited by 7
  9. 2025R1-ACT: Efficient Reasoning Model Safety Alignment by Activating Safety Knowledgecited by 6
  10. 2026What Matters For Safety Alignment?cited by 6
  11. 2025Reasoned Safety Alignment: Ensuring Jailbreak Defense via Answer-Then-Checkcited by 5
  12. 2025Off-Trajectory Reasoning: Can LLMs Collaborate on Reasoning Trajectory?cited by 5
  13. 2025When Models Outthink Their Safety: Unveiling and Mitigating Self-Jailbreak in Large Reasoning Modelscited by 5
  14. 2025Unveiling Trust in Multimodal Large Language Models: Evaluation, Analysis, and Mitigationcited by 4
  15. 2025AdvChain: Adversarial Chain-of-Thought Tuning for Robust Safety Alignment of Large Reasoning Modelscited by 4
  16. 2025Reasoning over Boundaries: Enhancing Specification Alignment via Test-time Deliberationcited by 3
  17. 2025ThinkPilot: Steering Reasoning Models via Automated Think-prefixes Optimizationcited by 3
  18. 2025Scaling Reinforcement Learning for Content Moderation with Large Language Modelscited by 3
  19. 2026How Does the Thinking Step Influence Model Safety? An Entropy-based Safety Reminder for LRMscited by 3
  20. 2026Mitigating Safety Tax via Distribution-Grounded Refinement in Large Reasoning Modelscited by 3
  21. 2025Does More Inference-Time Compute Really Help Robustness?cited by 2
  22. 2025ReasoningGuard: Safeguarding Large Reasoning Models with Inference-time Safety Aha Momentscited by 2
  23. 2025AURA: Affordance-Understanding and Risk-aware Alignment Technique for Large Language Modelscited by 2
  24. 2025Strong but Brittle: Simple Attacks Subvert Reasoning-based Safety Guardrailscited by 2
  25. 2025Beyond SFT: Reinforcement Learning for Safer Large Reasoning Models with Better Reasoning Abilitycited by 2
  26. 2025AI Security Beyond Core Domains: Resume Screening as a Case Study of Adversarial Vulnerabilities in Specialized LLM Applicationscited by 2
  27. 2026STAR-S: Improving Safety Alignment through Self-Taught Reasoning on Safety Rulescited by 2
  28. 2026Contrastive Reasoning Alignment: Reinforcement Learning from Hidden Representationscited by 2
  29. 2025HauntAttack: When Attack Follows Reasoning as a Shadowcited by 1
  30. 2025From Perception to Reasoning: Deep Thinking Empowers Multimodal Large Language Modelscited by 1
  31. 2026RAPO: Risk-Aware Preference Optimization for Generalizable Safe Reasoningcited by 1
  32. 2026Amplification Effects in Test-Time Reinforcement Learning: Safety and Reasoning Vulnerabilitiescited by 1
  33. 2026Deliberative Alignment is Deep, but Uncertainty Remains: Inference time safety improvement in reasoning via attribution of unsafe behavior to base modelcited by 1
  34. 2026Self-ReSET: Learning to Self-Recover from Unsafe Reasoning Trajectoriescited by 1
  35. 2025COSMO-RL: Towards Trustworthy LMRMs via Joint Safety and Stabilitycited by 0
  36. 2025Pluralistic Behavior Suite: Stress-Testing Multi-Turn Adherence to Custom Behavioral Policiescited by 0
  37. 2026How Does Prefix Matter in Reasoning Model Tuning?cited by 0
  38. 2026Dialectics of Alignment: Harnessing Unsafe Knowledge for Dynamic Safety Routingcited by 0

50 most cited of 58 All 58 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025Safety Evaluation and Enhancement of DeepSeek Models in Chinese Contexts
  2. 2025RAJ-PGA: Reasoning-Activated Jailbreak and Principle-Guided Alignment Framework for Large Reasoning Models
  3. 2025Challenges in Ensuring AI Safety in DeepSeek-R1 Models: The Shortcomings of Reinforcement Learning Strategies
  4. 2026Adaptive and Explicit safe: Triggering Latent Safety Awareness in Large Reasoning Models
  5. 2025H-CoT: Hijacking the Chain-of-Thought Safety Reasoning Mechanism to Jailbreak Large Reasoning Models, Including OpenAI o1/o3, DeepSeek-R1, and Gemini 2.0 Flash Thinking
  6. 2025100 Days After DeepSeek-R1: A Survey on Replication Studies and More Directions for Reasoning Language Models
  7. 2025Quantifying the Capability Boundary of DeepSeek Models: An Application-Driven Performance Analysis
  8. 2026Reasoning Structure Matters for Safety Alignment of Reasoning Models