I'm posting my new AI control ideas at Less Wrong here.

I will crosspost some of the most interesting ones, but most of them will be only on Less Wrong.

Let me know what you think.