Understanding Secret Leakage Risks in Code LLMs: A Tokenization Perspective
For developers and security practitioners, this work identifies a previously unknown vulnerability in code LLMs that increases secret leakage risks, highlighting a need for tokenizer redesign.
This study reveals that Byte-Pair Encoding tokenization causes a 'gibberish bias' in Code LLMs, making certain secrets easier to memorize due to token distribution shift between training data and secret data. The bias is supported by numerical evidence showing high character-level but low token-level entropy for these secrets.
Code secrets are sensitive assets for software developers, and their leakage poses significant cybersecurity risks. While the rapid development of AI code assistants powered by Code Large Language Models (CLLMs), CLLMs are shown to inadvertently leak such secrets due to a notorious memorization phenomenon. This study first reveals that Byte-Pair Encoding (BPE) tokenization leads to unexpected behavior of secret memorization, which we term as \textit{gibberish bias}. Specifically, we identified that some secrets are among the easiest for CLLMs to memorize. These secrets yield high character-level entropy, but low token-level entropy. Then, this paper supports the biased claim with numerical data. We identified that the roots of the bias are the token distribution shift between the CLLM training data and the secret data. We further discuss how gibberish bias manifests under the ``larger vocabulary'' trend. To conclude the paper, we discuss potential mitigation strategies and the broader implications on current tokenizer design.