AI Is 2x Better at Exploiting Smart Contract Flaws Than Catching Them, Binance Finds
For context, EVMbench is a benchmark that measures how well AI agents can detect, patch, and exploit high-severity smart contract vulnerabilities. It draws on 117 curated vulnerabilities from 40 audits Smart contracts hold billions in user funds across decentralized finance (DeFi). Their open-source code makes them ideal targets for automated probing. AI systems can scan thousands of contracts in minutes at marginal cost. The asymmetry is widening because attack costs are collapsing. Binance Research data shows AI-powered exploits average roughly $1.22 per contract, with that figure projected to fall another 22% every two months. “Hackens SSDLC Maturity Survey shows over 80% of developers now use AI in development, but fewer than 40% use AI for advanced testing — leaving the offense-defense gap structurally lopsided,” Binance Research added. The threat extends beyond static code. Analysts at TRM Labs have begun speculating that North Korean hackers are integrating AI into their reconnaissance and social engineering operations. The shift would help explain attacks like Drift, which involved weeks of targeted manipulation of sophisticated blockchain systems, a marked departure from North Koreas traditional reliance on basic private key compromises. When AI hits security there will be signs pic.twitter.com/aZrJDIz4Z4 — Kevin Kwok (@kevinakwok) April 30, 2026 AI Is Reshaping the Economics of Crypto