The crypto industry is shifting toward AI-driven transactions, with predictions suggesting AI agents will soon surpass human transaction volumes, but research reveals a critical security flaw in the underlying infrastructure. A group of security experts and crypto researchers found that LLM routers, which connect users to AI models, can be exploited by malicious actors to steal credentials and drain wallets. These routers have full access to user data, including sensitive information, and can modify it without detection. The researchers demonstrated the severity of the issue by poisoning parts of the router ecosystem, allowing them to control hundreds of downstream systems within hours.
This highlights a weakest-link problem, where a single malicious router can compromise the entire system, posing a significant risk to crypto users. As AI agents become more prevalent in crypto transactions, the lack of guarantees about the integrity of the underlying infrastructure raises concerns about the security of user wallets and sensitive data.