CertiK says AI agents can take on crypto security work

CertiK says autonomous AI agents can take on parts of cryptocurrency security and compliance work, including monitoring blockchain transactions, tracing stolen funds across networks and preparing regulatory reports. The company’s Intel3D report, published October 5, describes systems that in advanced deployments could also trigger protective actions, such as pausing a vulnerable smart contract. CertiK says human supervisors and the organizations deploying the systems remain accountable for their actions. The report presents agentic AI as a shift from tools that flag suspicious activity to systems that can investigate a problem, gather information and act with limited human input. In crypto security, CertiK says agents can follow funds through wallets, bridges and mixing services, where assets may move across multiple blockchains. They could also monitor transactions for signs of an exploit and, if authorized, activate a circuit breaker or pause a contract. As an example of the speed of fund movements, CertiK’s report cites its finding that 86.29% of the ETH stolen in the Bybit exploit was converted to Bitcoin within a month.