《OpenAI releases smart contract benchmark testing—what does it mean?》 (Author: @chaowxyz) OpenAI has launched EVMBench, a smart contract benchmark test to evaluate AI agents' abilities to understand, fix, and exploit smart contracts. It uses 120 high-risk vulnerabilities from 40 real-world projects as the test set, divided into three tasks: vulnerability discovery, code repair, and attack simulation. The author believes this is not just a contract capability assessment but an exam for agents' survival skills on-chain, pointing to fundamental issues of AI's autonomous actions, collaboration, and contract execution in the crypto environment. Read the full article: https://(x.com)/chaowxyz/status/2024358080910413973