The movement to legally ban "superintelligent" AI now has a British bill. On 8 September, Labour MP Alex Sobel presents the UK Artificial Superintelligence Security Bill in Parliament — a private member's bill drafted by the campaign group ControlAI. It would prohibit the development of superintelligent AI in Britain, require the government to monitor and restrict its precursors, and commit ministers to pursuing an international agreement to prohibit it.

Its most prominent backer is Geoffrey Hinton, the Nobel laureate whose work on neural networks underpins much of modern AI. "We would be very foolish to develop superintelligence now, when there is no scientific consensus it can be developed safely and controllably," Hinton said, warning that losing control of AI smarter than ourselves could be catastrophic and "could even lead to human extinction." The writer and broadcaster Stephen Fry, another signatory, framed it as a threat "of equal if not greater potential for existential harm than nuclear power." ControlAI says the bill has backing from more than 100 MPs and peers across parties — a claim worth noting as the campaign's own count rather than a tally of votes the bill would command.

This is the British front of a movement On The Wire has been tracking elsewhere. Days earlier, the US senator Bernie Sanders — with Representative Greg Casar — proposed banning superintelligent AI outright; the UK bill is the same instinct arriving through a different political system, and within days of it. It also lands the same week OpenAI's own chief scientist, Jakub Pachocki, warned that "no one is prepared" for the pace of machine intelligence and called for enforceable safety standards — a reminder that the case for restraint is no longer coming only from outside the labs.

The honest caveat is about mechanics, not merits. A private member's bill of this kind almost never becomes law: without government time and backing it will, on the odds, run out of parliamentary road, as most such bills do. Its function is agenda-setting — forcing a debate, putting names to a position, and testing how much cross-party appetite there is for hard limits rather than voluntary codes. Read that way, the "more than 100 MPs and peers" figure is the actual news: not that Britain is about to outlaw superintelligence, but that a sizeable bloc of legislators is now willing to attach its name to the idea.

There is a definitional problem underneath all of it, and the bill does not escape it. "Superintelligence" has no agreed technical meaning, no benchmark that switches from amber to red. A law that prohibits developing a thing no one can precisely define has to lean on "precursors" — and almost any frontier capability can be argued to be one. That is either the point (a deliberately broad brake) or the flaw (an unworkable line), depending on where you stand, and it is the argument the bill exists to provoke rather than settle.

What makes 8 September worth watching is not the bill's odds, which are long, but the coalition behind it: a Nobel laureate, a national-treasure broadcaster, a campaign group, and a triple-figure group of parliamentarians, arriving in the same fortnight as a US ban proposal and a frontier lab's own safety warning. The politics of AI has spent two years being made in California and Brussels. This week some of it is being made in Westminster.