What protections does SAFE have against nefarious nodes/software?

Continuing the discussion from Some random thoughts on updating autonomous networks: