HALURust introduces an innovative framework that exploits LLM hallucinations to detect security vulnerabilities in Rust programming language.

• Addresses the growing concern of security vulnerabilities in Rust despite its safety-focused design (442 vulnerabilities reported since 2018)
• Overcomes the challenge of limited vulnerability data by leveraging large language models in a novel way
• Demonstrates superior detection capabilities compared to existing vulnerability detection tools
• Provides a practical solution for identifying security weaknesses in an increasingly popular programming language

This research offers essential security tooling for organizations adopting Rust, helping to maintain its safety promises while addressing real-world vulnerability challenges.

HALURust: Exploiting Hallucinations of Large Language Models to Detect Vulnerabilities in Rust