Find out why growing numbers want action, and how ASVA drives safety.
You've probably noticed AI getting sharper every few months, but the pace is easy to underestimate. Systems that could barely string a sentence together a few years ago now write code, pass professional exams and outperform most people at tasks we assumed needed human judgement. And the explicit goal of the companies building it isn't a better chatbot: it's AI that matches, then exceeds, human intelligence across the board.
Getting there first is worth a fortune, so the leading labs are locked in a flat-out race with each other, and increasingly with China. Whoever ships the most capable system wins the market, the investment and the headlines. That pressure rewards speed over caution, and no one wants to be the lab that slows down while a rival races ahead.
Here's the part that isn't hype: no one, not even the people building these systems, knows how to reliably control an AI far smarter than us. And we're barely in control of the AI we already have. Today's models still get jailbroken, mislead their own testers and do things their makers didn't intend and can't fully explain. Scaling that up to something more capable than any human, before we've solved control, is why serious researchers put the worst case as high as human extinction. It's not a fringe view. Geoffrey Hinton, who won a Nobel Prize for the work behind modern AI, left Google specifically to warn about it, and in 2023 the heads of the leading labs signed a one-line statement putting AI extinction risk alongside pandemics and nuclear war.
The catch is that making these systems genuinely safe takes time: testing them, understanding how they work, holding back a release until you're sure. In a race, that time is exactly what gets cut. Safety is the first thing to go the moment a lab thinks it's falling behind, because caution is a disadvantage when your competitors won't share it.
You might assume governments have this covered. They don't. There is currently no binding law in the UK, or almost anywhere, that stops a company building and releasing a system like this. Most MPs have never had a single constituent raise AI safety with them, so it barely registers on the political radar as something voters care about, let alone something worth regulating.
None of this is hopeless. Researchers have real proposals: testing regimes, oversight rules, international agreements. And we've done hard things before: the world has largely held back nuclear, chemical and biological weapons through treaties and pressure. What's missing isn't ideas, it's will. Right now there's almost no political push to make any of it happen, and that's the one piece ordinary voters can actually change.
The AI Safety Voter Alliance is built on a simple idea: turn concern about AI into something politicians actually feel. We make AI safety a voting issue, the kind an MP can't afford to ignore.
We've researched all 650 MPs and graded them A to F on their AI safety record: how they've voted and what they've said publicly, backed by evidence rather than opinion. You can look up your own MP in seconds and see exactly where they stand.
Then we ask you to pledge that this grade will factor into how you vote, the way you might weigh the NHS or the environment. It takes about two minutes, and you don't need to be an expert, you're simply saying you're paying attention.
Single-issue voters punch far above their weight. An MP doesn't need tens of thousands of emails, just the knowledge that a defined group of their constituents is watching this issue and will remember at the ballot box. Right now that pressure is close to zero, so even a few hundred pledges per seat is a loud signal.
See where your MP stands, then add your name. It's the most concrete lever most of us have on how this technology gets handled, and it works best the earlier you use it.