OpenAI and Paradigm have launched EVMbench, a tool aimed at enhancing security for Ethereum Virtual Machine smart contracts by testing AI agents on their ability to detect, patch, and exploit vulnerabilities. The tool employs 120 vulnerabilities curated from real audits, such as those from Code4rena and Stripe’s new Tempo blockchain, highlighting the growing importance of AI in securing economically significant codebases amid the rising use of AI-driven stablecoin payments. Notably, the AI model GPT-5.3-Codex achieved a 72.2% success rate in the exploit testing mode, underscoring the potential dual role of AI as both a resource for security enhancement and a potential threat in smart contract auditing.
OpenAI introduces EVMbench to enhance Ethereum smart contract security
