# Think Before You Code: Dual Reasoning for the NLSafety-Utility Trade-Off in LLM Code Generation

arXiv: https://arxiv.org/abs/2604.12088 · HTML: https://mlsanity.com/paper/2604.12088/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 18 of its 46 references that are in the map.
- First posted on arXiv 13 Apr 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)

## Builds on (most cited first, 18 of 18)

- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://mlsanity.com/paper/2305.10601/) (2023)
- [Code Llama: Open Foundation Models for Code](https://mlsanity.com/paper/2308.12950/) (2023)
- [Least-to-Most Prompting Enables Complex Reasoning in Large Language Models](https://mlsanity.com/paper/2205.10625/) (2022)
- [Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation](https://mlsanity.com/paper/2305.01210/) (2023)
- [DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence](https://mlsanity.com/paper/2401.14196/) (2024)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [Safe RLHF: Safe Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2310.12773/) (2023)
- [Deliberative Alignment: Reasoning Enables Safer Language Models](https://mlsanity.com/paper/2412.16339/) (2024)
- [MultiPL-E: A Scalable and Extensible Approach to Benchmarking Neural Code Generation](https://mlsanity.com/paper/2208.08227/) (2022)
- [OASIS: Open Agent Social Interaction Simulations with One Million Agents](https://mlsanity.com/paper/2411.11581/) (2024)
- [STAIR: Improving Safety Alignment with Introspective Reasoning](https://mlsanity.com/paper/2502.02384/) (2025)
- [MTTM: Metamorphic Testing for Textual Content Moderation Software](https://mlsanity.com/paper/2302.05706/) (2023)
- [Synergy-of-Thoughts: Eliciting Efficient Reasoning in Hybrid Language Models](https://mlsanity.com/paper/2402.02563/) (2024)

## Abstract (arXiv)

Large language models (LLMs) for code generation are typically evaluated on functional correctness alone, overlooking whether generated code propagates harmful content embedded in the prompt. Prior work has shown that most Code LLMs reproduce offensive identifiers from injected renaming instructions without warning, yet existing approaches focus on detecting harmful content, neglecting functional correctness. Grounded in the Theory of Dual Channel Constraints (which states that code is a dual-channel medium combining an algorithmic (AL) channel for machine execution and a natural language (NL) channel for human communication, creating a unique safety-utility trade-off where a model must balance functional execution with responsible communication), we propose NLSafety-Utility Duality Score (SUDS), a metric that unifies code utility, safety adherence, and warning awareness into a single score across 12 ranked response scenarios, and Dual Reasoning (DR), a structured inference-time technique that requires an explicit safety audit and task-grounded code review before code generation. Evaluated on six LLMs across two benchmarks augmented with harmful keyword injections (820 and 2,135 samples), DR consistently achieves the highest SUDS across all models, improving mean SUDS by 1.32× to 3.42× over the baseline, while chain-of-thought prompting yields negligible safety gains and a safety-aware prompt provides only partial improvement. Further analysis reveals that DR's effectiveness scales with model capacity, that the one-shot exemplar primarily stabilizes output format for smaller models, and that structured reasoning cannot compensate for models with limited safety vocabularies.