Email: [email protected]
Many businesses believe that giving an AI agent read-only access mitigates security risks. However, indirect prompt injection means even read-only agents can exfiltrate sensitive data.
