UK Government Says AI Can't Be Turned Off Because It's 'More Reliable Than Politicians'

The Cabinet Office has officially declared that Britain cannot simply turn AI off. Not because the technology is too deeply embedded in critical infrastructure. Not because the economic fallout would be catastrophic. But because, and this is real, they’re worried it might work better than the current system.

The government’s position is that a ‘kill switch’ for dangerous AI is impractical. Which is true in the way that seat belts are impractical if you’re driving a car you’ve never tested and don’t fully understand. The logic here is bulletproof: we’ve built something potentially catastrophic into everything, so obviously we can’t remove it without breaking everything. Problem solved.

What’s genuinely impressive is the confidence. The same institution that cannot pass legislation without three elections’ worth of infighting has decided it’s too risky to maintain the ability to shut down artificial superintelligence. The same people who can’t agree on a budget are now arguing that emergency kill switches are philosophically unsound.

The real tell is buried in the reasoning. A kill switch, they say, would be unreliable. Unlike, presumably, the current British political system, which has never once failed to deliver on its promises or accidentally destroyed something important through sheer incompetence. The comparison is so unhinged it might actually be the most honest thing any government has said about AI all year.

So here we are. The UK has decided that the best way to manage existential risk is to make sure we can never actually manage it. Brilliant. Absolutely foolproof. What could go wrong.