Is Watermarking LLM-Generated Code Robust?

Tarun Suresh, Shubham Ugare, Gagandeep Singh, Sasa Misailovic

Keywords: large language models, watermarking, robustness

Abstract: We present the first study of the robustness of existing watermarking techniques on Python code generated by large language models. Although existing works showed that watermarking can be robust for natural language text, we show that it is easy to remove these watermarks on code by simple semantic-preserving transformations.

Submission Number: 196