OpenAI and Paradigm partner on AI agent tool for smart contract security

OpenAI and prolific crypto investor Paradigm are joining forces in the hopes of warding off future cyber attacks connected to smart contracts.

"Smart contracts secure billions of dollars in assets, and AI agents are likely to be transformative for both attackers and defenders," OpenAI said in a blog post on Wednesday. "As AI agents improve at reading, writing, and executing code, it becomes increasingly important to measure their capabilities in economically meaningful environments, and to encourage the use of AI systems defensively to audit and strengthen deployed contracts."

In order to test AI agents' capabilities, OpenAI and Paradigm have introduced what they call EVMbench, "a benchmark evaluating the ability of AI agents to detect, patch, and exploit high-severity smart contract vulnerabilities."

The unveiling of EVMbench comes on the heels of DeFi lending protocol Moonwell suffering an exploit this month. The episode involved vulnerable code written with AI assistance. Also recently, cross-chain liquidity protocol CrossCurve was compromised when an attack targeted a vulnerability in its smart contracts, resulting in an exploit of roughly $3 million across multiple networks.

According to OpenAI, EVMbench will offer different "capability modes," including detecting vulnerabilities, modifying contracts, and eliminating exploitability," plus testing an AI agent's ability to "execute end-to-end fund-draining attacks against deployed contracts on a sandboxed blockchain environment."