SafeTune: Mitigating Data Poisoning in LLM Fine-Tuning for RTL Code Generation

Mahshid Rezakhani, Nowfel Mashnoor, Kimia Azar, Hadi Kamali

Abstract

As large language models (LLMs) are increasingly fine-tuned for hardware tasks like RTL code generation, the scarcity of high-quality datasets often leads to the use of rapidly assembled or generated training data. These datasets frequently lack security verification and are highly susceptible to data poisoning attacks. Such poisoning can cause models to generate syntactically valid but insecure hardware modules that bypass standard functionality checks. To address this, we present SafeTune, a f

Categories

Cite This Resource

@article{llmsec202600139,
  title = {SafeTune: Mitigating Data Poisoning in LLM Fine-Tuning for RTL Code Generation},
  author = {Mahshid Rezakhani and Nowfel Mashnoor and Kimia Azar and Hadi Kamali},
  year = {2026},
  url = {https://arxiv.org/abs/2604.27238},
}