Why would it makes sense to remove that process, while introducing an incredible opaque decision-making process in its place, which totally bars anyone from knowing why they were excluded from accessing the internet? It even prevents wrongfully excluded individuals from receiving compensation.
For example, I could be cut off from the internet which I need to do my job. Say I'm unable to work for a week or two and then it's determined that I can access the internet because an error was made... Well, as far as I can tell, I'd be SOL. That doesn't seem right...
Worse still is that this seems about as technically competent as using an IP address to determine a person's location. Any serious threat vector, human or otherwise, will find other ways to access the internet or perpetuate their threat if they care to. If they're a serious threat, why wouldn't prison be a better solution than... Calling their ISP and banning them from the internet?
All of this seems very short-sighted, undemocratic, and naive.
And while the 'human or otherwise' phrase I used might seem odd (I know someone's dog isn't shit-posting on X), what I mean to say is something like... What if an LLM is posting from an unsuspecting person's computer and was placed there as a virus? Once it's cut off from that poor person's computer, it's very likely it will eventually or already be functioning from some other unsuspecting person's computer, server, or whatever other device. Their toaster. My point is that we live in an age where there are non-human agents causing harm online. The machine they operate from will not always be OpenAI's or Anthropics, and indeed, will probably rarely be so.
This was already the case with human actors, but it made much worse with the advent of AI-based agents.