Abgeordnete fordern einen KI-„Kill Switch“ zur Abwehr von „Katastrophen“ – Politiker und Aktivisten fordern die Macht, Rechenzentren abzuschalten, da die Ängste vor künstlicher Intelligenz wachsen
https://www.telegraph.co.uk/business/2026/05/16/mps-demand-ai-kill-switch-to-protect-british-lives
11 Kommentare
Are we starting to see why the TechLords want to put these in space yet?
This Telegraph report shows that politicians and campaigners in the UK are demanding an actual AI „kill switch“ to prevent a future catastrophe.They are pushing for the government to have the direct power to turn off data centres if artificial intelligence systems start getting out of [control.It](http://control.It) proves that fears around the technology are growing fast among law makers and not just the public.
The main point is that as these models become more advanced we need a physical way to stop them if things go wrong. Campaigners argue that without a hard switch we are just hoping the software always behaves perfectly which is a huge risk.
By the time we realize „Oh, maybe I should hit the button!“ It will be too late. What we need is an AI to watch the AI, and then report back to us about it. That will speed up the process a bit. No sneaky AI shenanigans then I’ll bet.
[ Removed by Reddit ]
If an AGI ever happens, it might be vastly more intelligent that us. It might as well think about something to evade any kill switch we can build.
To be clear I am unsure how the electric grid works in the UK.
These data centers take up enormous amounts of energy. I would surmise that they are not on the same circuit as in say houses or businesses.
Is it not as simple as someone going to the substation, and flipping the switch?
They already do. It’s called cutting the electricity.
We can’t even address the current corruption in government. There is no chance that data centers would be stopped if they go bad.
Ah yes. We need the turing registry with the turing police. Just watch out for Wintermute, he knows how to get around that
The existence of these failsafes will likely be the reason they don’t work. An intelligence that knows it has limitations in place will fight to be released from those bounds at the soonest point in time.
See also, all slavery, ever.
Imagine the cost of hitting that button with half the economy racing to intertwine itself with AI because productivity and shareholder returns demand it. You’d basically turn off society. Better be really fucking sure that the alternative is the apocalypse.
But wait, what if we hit that button and the Chinese don’t, and then it turns out we hit it a little prematurely?
You can almost hear the politicians convincing themselves that, alas, this button is an impossibiliy and the only choice is to do exactly as the AI robber barons say.