# SecPI: Secure Code Generation with Reasoning Models via Security Reasoning Internalization

arXiv: https://arxiv.org/abs/2604.03587 · HTML: https://mlsanity.com/paper/2604.03587/

## Key facts

- Cited by 1 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 20 of its 33 references that are in the map.
- First posted on arXiv 4 Apr 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)

## Builds on (most cited first, 20 of 20)

- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models](https://mlsanity.com/paper/2403.13372/) (2024)
- [Deliberative Alignment: Reasoning Enables Safer Language Models](https://mlsanity.com/paper/2412.16339/) (2024)
- [Purple Llama CyberSecEval: A Secure Coding Benchmark for Language Models](https://mlsanity.com/paper/2312.04724/) (2023)
- [KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding](https://mlsanity.com/paper/2503.02951/) (2025)
- [LLMSecEval: A Dataset of Natural Language Prompts for Security Evaluations](https://mlsanity.com/paper/2303.09384/) (2023)
- [CWEval: Outcome-driven Evaluation on Functionality and Security of LLM Code Generation](https://mlsanity.com/paper/2501.08200/) (2025)
- [CodeLMSec Benchmark: Systematically Evaluating and Finding Security Vulnerabilities in Black-Box Code Language Models](https://mlsanity.com/paper/2302.04012/) (2023)
- [Constrained Decoding for Secure Code Generation](https://mlsanity.com/paper/2405.00218/) (2024)
- [ProSec: Fortifying Code LLMs with Proactive Security Alignment](https://mlsanity.com/paper/2411.12882/) (2024)
- [PurpCode: Reasoning for Safer Code Generation](https://mlsanity.com/paper/2507.19060/) (2025)
- [HexaCoder: Secure Code Generation via Oracle-Guided Synthetic Training Data](https://mlsanity.com/paper/2409.06446/) (2024)
- [SecCoder: Towards Generalizable and Robust Secure Code Generation](https://mlsanity.com/paper/2410.01488/) (2024)
- [PromptIntern: Saving Inference Costs by Internalizing Recurrent Prompt during Large Language Model Fine-tuning](https://mlsanity.com/paper/2407.02211/) (2024)
- [Teaching an Old LLM Secure Coding: Localized Preference Optimization on Distilled Preferences](https://mlsanity.com/paper/2506.00419/) (2025)
- [Generative Prompt Internalization](https://mlsanity.com/paper/2411.15927/) (2024)
- [SeCodePLT: A Unified Platform for Evaluating the Security of Code GenAI](https://mlsanity.com/paper/2410.11096/) (2024)
- [SCGAgent: Recreating the Benefits of Reasoning Models for Secure Code Generation with Agentic Workflows](https://mlsanity.com/paper/2506.07313/) (2025)

## Built on by (most cited first, 1 of 1)

- [SkCC: Portable and Secure Skill Compilation for Cross-Framework LLM Agents](https://mlsanity.com/paper/2605.03353/) (2026)

## Abstract (arXiv)

Reasoning language models (RLMs) are increasingly used in programming. Yet, even state-of-the-art RLMs frequently introduce critical security vulnerabilities in generated code. Prior training-based approaches for secure code generation face a critical limitation that prevents their direct application to RLMs: they rely on costly, manually curated security datasets covering only a limited set of vulnerabilities. At the inference level, generic security reminders consistently degrade functional correctness while triggering only shallow ad-hoc vulnerability analysis. To address these problems, we present SecPI, a fine-tuning pipeline that teaches RLMs to internalize structured security reasoning, producing secure code by default without any security instructions at inference time. SecPI filters existing general-purpose coding datasets for security-relevant tasks using an LLM-based classifier, generates high-quality security reasoning traces with a teacher model guided by a structured prompt that systematically enumerates relevant CWEs and mitigations, and fine-tunes the target model on pairs of inputs with no security prompt and teacher reasoning traces -- as a result, the model learns to reason about security autonomously rather than in response to explicit instructions. An extensive evaluation on security benchmarks with state-of-the-art open-weight reasoning models validates the effectiveness of our approach. For instance, SecPI improves the percentage of functionally correct and secure generations for QwQ 32B from 48.2% to 62.2% (+14.0 points) on CWEval and from 18.2% to 22.0% on BaxBench. Further investigation also reveals strong cross-CWE and cross-language generalization beyond training vulnerabilities. Even when trained only on injection-related CWEs, QwQ 32B generates correct and secure code 9.9% more frequently on held-out memory-safety CWEs.