Actually if you install a llm machine in local, you can already train it with your blogs, writings and such to replicate your writing style, it's not future, it's now already...

I somehow gave up on being secure giving my own data and Id to exchanges and such, simply because the hospital where I've gone has been hacked, all data taken and sold,same for an other clinic, and a betting site so probably my data is well spread in the deep web...

As for how to secure there's a single user, it's a milion dollar question, iris scan I don't like it I'm pretty sure can be faked somehow, check user DNA hashed every time like with saliva would maybe do it, I can hardly see faking saliva