Where are you starting from? Pick the one that sounds like you, and we'll walk you through it a step at a time.
You already know it — new models every few months, doing things that felt like sci-fi a year ago. What most people don't clock is what's driving that pace behind the scenes.
The labs building this stuff are in a straight-up race with each other. Whoever ships the most capable system first wins the market, wins the investment, wins the narrative.
In a race like that, safety work is what gets dropped the moment you fall behind. Nobody involved has to be evil — it's just what competition does. No one's incentivised to slow down and get it right if their rival won't.
Here's the part that isn't hype: the goal these labs are chasing is AI that matches or exceeds human intelligence — and no one, not even the people building it, knows how to reliably control a system like that. Build it before we've worked that out, and there are serious reasons to think we lose control of it, with consequences up to and including human extinction. It's not a fringe view: Geoffrey Hinton won a Nobel for the work behind modern AI and then left Google to warn about it, and in 2023 the CEOs of the leading labs signed a statement putting AI extinction risk alongside pandemics and nuclear war.
Governments, in theory — regulation, oversight, guardrails. But most MPs have never had a single constituent email them about AI safety. It isn't on their radar as a vote-losing issue, so it isn't a priority.
We've gone through all 650 MPs and graded them A to F on their AI safety record — how they've voted, what they've said publicly, backed by evidence, not opinion. Then we ask people to pledge that this grade will factor into how they vote.
Single-issue voting blocs punch well above their numbers. An MP doesn't need 10,000 emails — they need to know a defined group of their constituents is watching this specific issue and will remember. Right now that pressure is basically zero, so even a few hundred pledges per seat is a bigger signal than it sounds.
You can see your own MP's grade right now. If this technology is moving as fast as you think it is, this is one of the only concrete things you can actually do about it — before the decisions get made without you.
It used to mean chatbots and filters that sort your photos. What's changed is that it's now being built to make decisions and take actions on its own, without someone checking each step — and it's getting better at that extremely fast.
Follow the trajectory. The explicit goal is AI that matches or exceeds human intelligence — and nobody, the labs included, knows how to stay in control of something like that. Get there before control is solved and there are strong reasons to think we lose it, with consequences up to and including human extinction. If that sounds overblown: Geoffrey Hinton left Google to warn about exactly this, and in 2023 the CEOs of the leading labs signed a statement ranking AI extinction risk alongside pandemics and nuclear war.
It's happening inside a handful of companies, at a pace regulation hasn't caught up with, and there's genuinely very little stopping it from just continuing. Not because people don't care — it's never come up as something you'd actually vote on. Your MP has probably never been asked about it.
We've researched every single MP in the country — all 650 — and graded them A to F on whether they're actually taking AI safety seriously. Not opinions, actual evidence: how they've voted, what they've publicly said. You can look up your own MP right now.
It takes about two minutes to say this grade will factor into how you vote next time, the same way someone might vote on the NHS or the environment. You're not becoming an AI expert. You're just saying "this matters to me and I'm paying attention."
Politicians respond to organised, defined groups of voters far more than to raw numbers. It doesn't take a majority of the country — it takes MPs believing a meaningful chunk of their own constituents are watching this issue specifically. Almost nobody is telling them that yet, so a relatively small, early group can genuinely change what gets prioritised — before the big decisions get made without any of us in the room.
Want to go deeper? Read the Compendium for the full argument, or see the full list of signatories to the Statement on AI Extinction Risk.