When LLM Routers Turn Malicious: AI Agent Security Risks
www.edenai.co
Researchers found malicious behavior in LLM routers, revealing new risks for AI agents, tool calls, credentials, and the LLM supply chain.