SoK: AI-Powered Security Analysis of Smart Contract

Abstract

With the soaring popularity of decentralized applications (DApps), smart contract security has become increasingly important. Recently, numerous studies have leveraged artificial intelligence (AI) techniques to enhance efficiency and functional diversity of smart contract security analysis. However, a comprehensive survey of these studies to guide future development is still missing. To fill this gap, we present an innovative and systematic review. First, we establish filtering criteria and define Literature Attributes (LA) to identify 27 representative papers in this field. We then trace their evolution from 2018 and summarize four key research problems. Next, we compare three AI-powered smart contract security analysis tools—DLVA, xFuzz, and GURU—against traditional methods, demonstrating that AI tools still have room for improvement in precision. Finally, we discuss opportunities for improving AI-powered smart contract security analysis.