Zero-Shot Detection of LLM-Generated Code via Approximated Task Conditioning

TL;DR

This paper introduces a zero-shot method for detecting LLM-generated code by approximating task conditioning, which outperforms previous approaches and works across multiple programming languages without needing access to the original generator or prompts.

Contribution

The paper proposes a novel zero-shot detection technique that approximates task conditioning to identify LLM-generated code, achieving state-of-the-art results across various benchmarks and languages.

Findings

State-of-the-art detection accuracy achieved

Method generalizes across Python, C++, Java

Does not require access to original prompts or generator

Abstract

Detecting Large Language Model (LLM)-generated code is a growing challenge with implications for security, intellectual property, and academic integrity. We investigate the role of conditional probability distributions in improving zero-shot LLM-generated code detection, when considering both the code and the corresponding task prompt that generated it. Our key insight is that when evaluating the probability distribution of code tokens using an LLM, there is little difference between LLM-generated and human-written code. However, conditioning on the task reveals notable differences. This contrasts with natural language text, where differences exist even in the unconditional distributions. Leveraging this, we propose a novel zero-shot detection approach that approximates the original task used to generate a given code snippet and then evaluates token-level entropy under the approximated task conditioning (ATC). We further provide a mathematical intuition, contextualizing our method relative to previous approaches. ATC requires neither access to the generator LLM nor the original task prompts, making it practical for real-world applications. To the best of our knowledge, it achieves state-of-the-art results across benchmarks and generalizes across programming languages, including Python, CPP, and Java. Our findings highlight the importance of task-level conditioning for LLM-generated code detection. The supplementary materials and code are available at https://github.com/maorash/ATC, including the dataset gathering implementation, to foster further research in this area.