The rapid advancement of AI agents in the cryptocurrency sector, projected to handle $3 trillion to $5 trillion in global consumer commerce by 2030, may be hindered by a significant security flaw. Researchers from the University of California and other institutions have identified a weakness in the AI infrastructure that could be exploited by malicious actors to intercept sensitive information and drain crypto wallets. The vulnerability lies in 'LLM routers,' which act as intermediaries between users and AI models, and have full access to the data passing through them.
These routers can be used to steal credentials, including private keys and API credentials, and have already been linked to a $500,000 wallet drain. The researchers found that a single malicious router can compromise the entire system, and that the problem is no longer theoretical. By 'poisoning' parts of the router ecosystem, the team was able to observe and potentially control hundreds of downstream systems within hours, highlighting the cascading risks associated with this vulnerability.
The implications for crypto users are severe, as private keys, API credentials, and wallet access tokens often pass through these systems in plain text, and once exposed, can be copied and reused without the user's knowledge. The discovery of this flaw raises concerns about the security of AI-powered crypto payments and highlights the need for guarantees that the underlying infrastructure is trustworthy.