From a security perspective, it’s axiomatic that you can’t make a client behave in any particular way; we just assume the client is hostile.