# Reasoning over Boundaries: Enhancing Specification Alignment via Test-time Deliberation

arXiv: https://arxiv.org/abs/2509.14760 · HTML: https://mlsanity.com/paper/2509.14760/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 43 of its 74 references that are in the map.
- First posted on arXiv 18 Sep 2025.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM alignment / RLHF / preference optimisation / [Safety alignment and jailbreak attacks](https://mlsanity.com/alignment/safety-alignment-and-jailbreak-attacks/)

## Builds on (most cited first, 30 of 43)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [s1: Simple test-time scaling](https://mlsanity.com/paper/2501.19393/) (2025)
- [Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators](https://mlsanity.com/paper/2404.04475/) (2024)
- [DeepSeek-V3 Technical Report](https://mlsanity.com/paper/2412.19437/) (2024)
- [Safe RLHF: Safe Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2310.12773/) (2023)
- [BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructions](https://mlsanity.com/paper/2406.15877/) (2024)
- [The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning](https://mlsanity.com/paper/2403.03218/) (2024)
- [From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline](https://mlsanity.com/paper/2406.11939/) (2024)
- [Deliberative Alignment: Reasoning Enables Safer Language Models](https://mlsanity.com/paper/2412.16339/) (2024)
- [SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language Models](https://mlsanity.com/paper/2402.05044/) (2024)
- [Mitigating the Alignment Tax of RLHF](https://mlsanity.com/paper/2309.06256/) (2023)
- [InFoBench: Evaluating Instruction Following Ability in Large Language Models](https://mlsanity.com/paper/2401.03601/) (2024)
- [SafeChain: Safety of Language Models with Long Chain-of-Thought Reasoning Capabilities](https://mlsanity.com/paper/2502.12025/) (2025)
- [Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable](https://mlsanity.com/paper/2503.00555/) (2025)
- [The Hidden Risks of Large Reasoning Models: A Safety Assessment of R1](https://mlsanity.com/paper/2502.12659/) (2025)
- [FOFO: A Benchmark to Evaluate LLMs' Format-Following Capability](https://mlsanity.com/paper/2402.18667/) (2024)
- [STAIR: Improving Safety Alignment with Introspective Reasoning](https://mlsanity.com/paper/2502.02384/) (2025)
- [STAR-1: Safer Alignment of Reasoning LLMs with 1K Data](https://mlsanity.com/paper/2504.01903/) (2025)
- [Effectively Controlling Reasoning Models through Thinking Intervention](https://mlsanity.com/paper/2503.24370/) (2025)
- [TreeBoN: Enhancing Inference-Time Alignment with Speculative Tree-Search and Best-of-N Sampling](https://mlsanity.com/paper/2410.16033/) (2024)
- [RealSafe-R1: Safety-Aligned DeepSeek-R1 without Compromising Reasoning Capability](https://mlsanity.com/paper/2504.10081/) (2025)
- [Explainable and Interpretable Multimodal Large Language Models: A Comprehensive Survey](https://mlsanity.com/paper/2412.02104/) (2024)
- [SETS: Leveraging Self-Verification and Self-Correction for Improved Test-Time Scaling](https://mlsanity.com/paper/2501.19306/) (2025)
- [AGENTIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios](https://mlsanity.com/paper/2505.16944/) (2025)

## Built on by (most cited first, 4 of 4)

- [Multimodal Policy Internalization for Conversational Agents](https://mlsanity.com/paper/2510.09474/) (2025)
- [The Compliance Gap: Why AI Systems Promise to Follow Process Instructions but Don't](https://mlsanity.com/paper/2605.01771/) (2026)
- [Analyzing and Internalizing Complex Policy Documents for LLM Agents](https://mlsanity.com/paper/2510.11588/) (2025)
- [User-Aware Active Knowledge Acquisition for Emotional Support Dialogue](https://mlsanity.com/paper/2605.29715/) (2026)

## Abstract (arXiv)

Large language models (LLMs) are increasingly applied in diverse real-world scenarios, each governed by bespoke behavioral and safety specifications (spec) custom-tailored by users or organizations. These spec, categorized into safety-spec and behavioral-spec, vary across scenarios and evolve with changing preferences and requirements. We formalize this challenge as specification alignment, focusing on LLMs' ability to follow dynamic, scenario-specific spec from both behavioral and safety perspectives. To address this challenge, we propose Align3, a lightweight method that employs Test-Time Deliberation (TTD) with hierarchical reflection and revision to reason over the specification boundaries. We further present SpecBench, a unified benchmark for measuring specification alignment, covering 5 scenarios, 103 spec, and 1,500 prompts. Experiments on 15 reasoning and 18 instruct models with several TTD methods, including Self-Refine, TPO, and MoreThink, yield three key findings: (i) test-time deliberation enhances specification alignment; (ii) Align3 advances the safety-helpfulness trade-off frontier with minimal overhead; (iii) SpecBench effectively reveals alignment gaps. These results highlight the potential of test-time deliberation as an effective strategy for reasoning over the real-world specification boundaries. Our code and resources are available at https://github.com/zzzhr97/SpecBench.