From Hard Refusals to Safe-Completions: Toward Output-Centric Safety Training

• URL: http://arxiv.org/abs/2508.09224v1
• Date: Tue, 12 Aug 2025 00:18:23 GMT
• Title: From Hard Refusals to Safe-Completions: Toward Output-Centric Safety Training
• Abstract summary: We propose safe-completions: a safety-training approach that centers on the safety of the assistant's output, rather than a binary classification of the user's intent.<n>We find that safe-completion training improves safety (especially on dual-use prompts), reduces the severity of residual safety failures, and substantially increases model helpfulness.
• Score: 14.489448208361175
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models used in ChatGPT have traditionally been trained to learn a refusal boundary: depending on the user's intent, the model is taught to either fully comply or outright refuse. While this is a strong mitigation for explicitly malicious prompts, focusing safety training on refusals can lead to brittleness for prompts with obscured user intent. Binary refusal boundaries are especially ill-suited for dual-use cases (such as biology or cybersecurity), where a user request can be answered safely at a high level, but in some cases can lead to malicious uplift if sufficiently detailed or actionable. As an alternative, we propose safe-completions: a safety-training approach that centers on the safety of the assistant's output, rather than a binary classification of the user's intent. Safe-completions seek to maximize helpfulness within the safety policy's constraints. We incorporated this approach into GPT-5 and find that across both production comparisons and internally controlled experiments, safe-completion training improves safety (especially on dual-use prompts), reduces the severity of residual safety failures, and substantially increases model helpfulness.

Related papers

• OpenSafeIntent: Evaluating Intent-Calibrated Safe Completion Across Dual-Use Prompt Sets [6.985945816612278]
  We introduce OpenSafeIntent, a benchmark of controlled prompt-sets that vary intent while holding the underlying task fixed.<n>This design lets us evaluate whether models calibrate assistance across intent shifts, rather than merely appearing safe on average.<n>Across a broad model suite, we find that prompt-level safety hides important failures.
  arXiv  Detail & Related papers  (2026-07-02T11:14:52Z)
• Safety Assessment in Reinforcement Learning via Model Predictive Control [3.244287913152012]
  We propose to leverage reversibility as a method for preventing safety issues throughout the training process.<n>Our method uses model-predictive path integral control to check the safety of an action proposed by a learned policy throughout training.
  arXiv  Detail & Related papers  (2025-10-23T19:31:18Z)
• Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Models [93.5740266114488]
  Constructive Safety Alignment (CSA) protects against malicious misuse while actively guiding vulnerable users toward safe and helpful results.<n>Oy1 achieves state-of-the-art safety among open models while retaining high general capabilities.<n>We release Oy1, code, and the benchmark to support responsible, user-centered AI.
  arXiv  Detail & Related papers  (2025-09-02T03:04:27Z)
• AlphaAlign: Incentivizing Safety Alignment with Extremely Simplified Reinforcement Learning [21.399086197886202]
  Large language models (LLMs) possess latent safety understanding from their vast pretraining data.<n>We propose textbfAlphaAlign, a pure reinforcement learning framework with verifiable safety reward.<n>This allows the model to develop proactive safety reasoning capabilities without depending on supervised safety-specific reasoning data.
  arXiv  Detail & Related papers  (2025-07-20T14:47:03Z)
• Reasoning as an Adaptive Defense for Safety [31.00328416755368]
  We build a recipe called $textitTARS$ (Training Adaptive Reasoners for Safety)<n>We train models to reason about safety using chain-of-thought traces and a reward signal that balances safety with task completion.<n>Our work provides an effective, open recipe for training LLMs against jailbreaks and harmful requests by reasoning per prompt.
  arXiv  Detail & Related papers  (2025-07-01T17:20:04Z)
• RSafe: Incentivizing proactive reasoning to build robust and adaptive LLM safeguards [55.76285458905577]
  Large Language Models (LLMs) continue to exhibit vulnerabilities despite deliberate safety alignment efforts.<n>To safeguard against the risk of policy-violating content, system-level moderation via external guard models has emerged as a prevalent mitigation strategy.<n>We propose RSafe, an adaptive reasoning-based safeguard that conducts guided safety reasoning to provide robust protection within the scope of specified safety policies.
  arXiv  Detail & Related papers  (2025-06-09T13:20:04Z)
• Shape it Up! Restoring LLM Safety during Finetuning [66.46166656543761]
  Finetuning large language models (LLMs) enables user-specific customization but introduces critical safety risks.<n>We propose dynamic safety shaping (DSS), a framework that uses fine-grained safety signals to reinforce learning from safe segments of a response while suppressing unsafe content.<n>We present STAR-DSS, guided by STAR scores, that robustly mitigates finetuning risks and delivers substantial safety improvements across diverse threats, datasets, and model families.
  arXiv  Detail & Related papers  (2025-05-22T18:05:16Z)
• SAFER: Advancing Safety Alignment via Efficient Ex-Ante Reasoning [51.78514648677898]
  We propose SAFER, a framework for Safety Alignment via eFficient Ex-Ante Reasoning.<n>Our approach instantiates structured Ex-Ante reasoning through initial assessment, rule verification, and path calibration.<n> Experiments on multiple open-source LLMs demonstrate that SAFER significantly enhances safety performance while maintaining helpfulness and response efficiency.
  arXiv  Detail & Related papers  (2025-04-03T16:07:38Z)
• Safe Vision-Language Models via Unsafe Weights Manipulation [75.04426753720551]
  We revise safety evaluation by introducing Safe-Ground, a new set of metrics that evaluate safety at different levels of granularity.<n>We take a different direction and explore whether it is possible to make a model safer without training, introducing Unsafe Weights Manipulation (UWM)<n>UWM uses a calibration set of safe and unsafe instances to compare activations between safe and unsafe content, identifying the most important parameters for processing the latter.
  arXiv  Detail & Related papers  (2025-03-14T17:00:22Z)
• Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing [12.986006070964772]
  Safety alignment is an essential research topic for real-world AI applications.<n>Our study first identified the difficulty of eliminating such vulnerabilities without sacrificing the model's helpfulness.<n>Our method could enhance the model's helpfulness while maintaining safety, thus improving the trade-off-front.
  arXiv  Detail & Related papers  (2025-02-04T09:31:54Z)
• Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training [67.30423823744506]
  We introduce a novel approach, Decoupled Refusal Training (DeRTa), designed to empower LLMs to refuse compliance to harmful prompts at any response position.<n>DeRTa incorporates two novel components: (1) Maximum Likelihood Estimation with Harmful Response Prefix, which trains models to recognize and avoid unsafe content by appending a segment of harmful response to the beginning of a safe response, and (2) Reinforced Transition Optimization (RTO), which equips models with the ability to transition from potential harm to safety refusal consistently throughout the harmful response sequence.
  arXiv  Detail & Related papers  (2024-07-12T09:36:33Z)
• Safe Reinforcement Learning with Learned Non-Markovian Safety Constraints [15.904640266226023]
  We design a safety model that performs credit assignment to assess contributions of partial state-action trajectories on safety. We derive an effective algorithm for optimizing a safe policy using the learned safety model. We devise a method to dynamically adapt the tradeoff coefficient between safety reward and safety compliance.
  arXiv  Detail & Related papers  (2024-05-05T17:27:22Z)
• Provable Safe Reinforcement Learning with Binary Feedback [62.257383728544006]
  We consider the problem of provable safe RL when given access to an offline oracle providing binary feedback on the safety of state, action pairs. We provide a novel meta algorithm, SABRE, which can be applied to any MDP setting given access to a blackbox PAC RL algorithm for that setting.
  arXiv  Detail & Related papers  (2022-10-26T05:37:51Z)
• Enforcing Hard Constraints with Soft Barriers: Safe Reinforcement Learning in Unknown Stochastic Environments [84.3830478851369]
  We propose a safe reinforcement learning approach that can jointly learn the environment and optimize the control policy. Our approach can effectively enforce hard safety constraints and significantly outperform CMDP-based baseline methods in system safe rate measured via simulations.
  arXiv  Detail & Related papers  (2022-09-29T20:49:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.