# Can LLMs Obfuscate Code? A Systematic Analysis of Large Language Models into Assembly Code Obfuscation

arXiv: https://arxiv.org/abs/2412.16135 · HTML: https://mlsanity.com/paper/2412.16135/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Dec 2024.

## Built on by (most cited first, 6 of 6)

- [Large Language Model (LLM) for Software Security: Code Analysis, Malware Analysis, Reverse Engineering](https://mlsanity.com/paper/2504.07137/) (2025)
- [What Breaks When LLMs Code? Characterizing Operational Safety Failures of Agentic Code Assistants](https://mlsanity.com/paper/2605.30777/) (2026)
- [Cybersecurity is the True Frontier for Generative AI Success or Failure](https://mlsanity.com/paper/2606.28929/) (2026)
- [ObfusQAte: A Proposed Framework to Evaluate LLM Robustness on Obfuscated Factual Question Answering](https://mlsanity.com/paper/2508.07321/) (2025)
- [Don't Judge a Book by its Cover: Testing LLMs' Robustness Under Logical Obfuscation](https://mlsanity.com/paper/2602.01132/) (2026)
- [SoK: AI-Augmented Binary Reversing](https://mlsanity.com/paper/2606.17398/) (2026)

## Abstract (arXiv)

Malware authors often employ code obfuscations to make their malware harder to detect. Existing tools for generating obfuscated code often require access to the original source code (e.g., C++ or Java), and adding new obfuscations is a non-trivial, labor-intensive process. In this study, we ask the following question: Can Large Language Models (LLMs) potentially generate a new obfuscated assembly code? If so, this poses a risk to anti-virus engines and potentially increases the flexibility of attackers to create new obfuscation patterns. We answer this in the affirmative by developing the MetamorphASM benchmark comprising MetamorphASM Dataset (MAD) along with three code obfuscation techniques: dead code, register substitution, and control flow change. The MetamorphASM systematically evaluates the ability of LLMs to generate and analyze obfuscated code using MAD, which contains 328,200 obfuscated assembly code samples. We release this dataset and analyze the success rate of various LLMs (e.g., GPT-3.5/4, GPT-4o-mini, Starcoder, CodeGemma, CodeLlama, CodeT5, and LLaMA 3.1) in generating obfuscated assembly code. The evaluation was performed using established information-theoretic metrics and manual human review to ensure correctness and provide the foundation for researchers to study and develop remediations to this risk.