You can’t scroll for long at the moment without another AI doom headline. Not just the job displacement doom, but the existential threat of AI.

Recently, they’ve started to quote former/current Anthropic employees, warning that the thing they built is/could be spiralling out of control.

People often take those warnings at face value. Conversations with friends, family, across the dinner table, all become a horror story and something else to worry about. It’s hard to argue otherwise, it almost sounds cavalier to do so.

A clue has emerged this week that may help. Anthropic (the company behind Claude) is accused of quietly building a machine that benefits from AI looking dangerous.

Here’s the argument, stripped down.

Anthropic is growing fast. Money flows into the AI-safety organisations around it. Those organisations warn that AI is getting scarier. The media amplifies it. Politicians respond by demanding more regulation. And of course the company that’s already deep in “AI safety“ is best placed to handle the new rules. Repeat that as a loop, and you have a self-serving situation.

Strip away the tech angle and it’s just regulatory capture, which isn’t new or exotic. You’ll recognise the shape from two other places.

Politics runs on it constantly. Manufacture the fear, then arrive selling the solution. Terror threats, crime waves, immigration panics: stoke the worry, then hand someone the mandate to “fix“ it.

Finance has run a similar play for decades. Heavier regulation gets sold as consumer protection, but mostly it just prices out anyone too small to afford the compliance department. The rules survive, the small players don’t.

The rules that “protect“ the public also protect the incumbent’s market share. That’s not a conspiracy theory, it’s just how power works when you’re already sitting at the top. Grow a business big enough and quick enough, and you can afford the inspectors and paperwork that will just bankrupt the smaller competition.

What makes the AI version interesting is the doom loop baked into it. The scarier AI sounds, the more funding and authority flows to the people already positioned as the “adults in the room”.

Fear isn’t just a side effect here, it’s useful.

I don’t think this means Anthropic sat in a room and plotted it. Maybe I’m wrong. Regulatory capture rarely needs a conspiracy. It just needs incentives that all point the same direction, and people who are genuinely convinced they’re doing the right thing, while also, conveniently, entrenching their own position.

That’s why it’s a more unsettling version. Not villains, just an ecosystem that rewards fear whether anyone intended it to or not.

Watch who’s funding the people telling you something is dangerous.

Not because they’re necessarily lying. But because “we need more oversight“ said by whoever’s best placed to survive that oversight should always get a second look.

Weirdly, Trump has come out on the side of no regulation. Maybe he’s spotted it too and it jealous of the power and money in AI, and would rather cause chaos than see them win.

So I’m not on board with team “AI will kill us all“ or team “it’s all hype”. But I am on board with team “follow the money before you follow the fear”.

As ever, we’ll have to see how it plays out.