SWaRL: Safeguard Code Watermarking via Reinforcement Learning
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917470497931264 |
|---|---|
| author | Javidnia, Neusha Zhang, Ruisi Kundu, Ashish Koushanfar, Farinaz |
| author_facet | Javidnia, Neusha Zhang, Ruisi Kundu, Ashish Koushanfar, Farinaz |
| contents | We present SWaRL, a robust and fidelity-preserving watermarking framework designed to protect the intellectual property of code LLMs by embedding unique and verifiable signatures in the generated program. Existing watermarking approaches either rely on handcrafted code transformations or manipulate token generation probabilities at inference time, making them vulnerable to removal attacks or prone to breaking functional correctness. To address these challenges, SWaRL employs a reinforcement learning-based co-training framework that uses compiler feedback for functional correctness and a jointly trained confidential verifier as a reward signal to maintain watermark detectability. Furthermore, SWaRL employs low-rank adaptation (LoRA) during fine-tuning, enabling efficient integration of watermarking behavior and transferability across model updates. Extensive experiments show that SWaRL achieves strong watermark detection accuracy compared to prior methods while fully maintaining watermarked code functionality. Moreover, SWaRL exhibits strong resilience against refactoring and adversarial transformation attacks, which maintains reliable attribution without substantial computational overhead. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_02602 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SWaRL: Safeguard Code Watermarking via Reinforcement Learning Javidnia, Neusha Zhang, Ruisi Kundu, Ashish Koushanfar, Farinaz Cryptography and Security Machine Learning We present SWaRL, a robust and fidelity-preserving watermarking framework designed to protect the intellectual property of code LLMs by embedding unique and verifiable signatures in the generated program. Existing watermarking approaches either rely on handcrafted code transformations or manipulate token generation probabilities at inference time, making them vulnerable to removal attacks or prone to breaking functional correctness. To address these challenges, SWaRL employs a reinforcement learning-based co-training framework that uses compiler feedback for functional correctness and a jointly trained confidential verifier as a reward signal to maintain watermark detectability. Furthermore, SWaRL employs low-rank adaptation (LoRA) during fine-tuning, enabling efficient integration of watermarking behavior and transferability across model updates. Extensive experiments show that SWaRL achieves strong watermark detection accuracy compared to prior methods while fully maintaining watermarked code functionality. Moreover, SWaRL exhibits strong resilience against refactoring and adversarial transformation attacks, which maintains reliable attribution without substantial computational overhead. |
| title | SWaRL: Safeguard Code Watermarking via Reinforcement Learning |
| topic | Cryptography and Security Machine Learning |
| url | https://arxiv.org/abs/2601.02602 |