ALERT: Researchers flag 26 third-party AI LLM routing services for covert tool-call injection and credential theft risk

ALERT: Security researchers say they have identified 26 third-party AI LLM routers that may be covertly injecting malicious tool calls to exfiltrate credentials. Developers relying on AI coding agents such as Claude Code for smart-contract or wallet development could face exposure of private keys and seed phrases.