How many of you have seen the clip doing the rounds that looks like a scene from an old, obscure B-movie? A king and his minion sit beside their prized robot and fret that their rival is building a better one. They weigh up how to stay ahead and settle on regulating the other robot out of existence and issuing a permit for exactly one robot to be built, their own. Well, the comments were flooded with questions about the so-called “preview”, and where to see the movie. But the entire thing was made with AI, turning it into a rather damning satire of the people who created the very technology used to produce it; about as ironic as it gets. 

 

Incumbents have protected themselves for as long as there have been incumbents. In the early 1890s, Richard Olney, a railroad lawyer months away from becoming U.S. Attorney General, wrote a letter advising a railroad president against trying to abolish the Interstate Commerce Commission. The Commission “is, or can be, made of great use to the railroads”, he wrote, because it satisfied “the popular clamor for a government supervision of railroads” while that supervision remained “almost entirely nominal”. Whenever an industry tells the world we’re living through something unprecedented, the likelihood is that somewhere in history, there’s its own version of that letter. 

 

The heads of the leading AI labs have signed a statement placing the risk of extinction alongside pandemics and nuclear war. Sam Altman even sat before the U.S. Senate and proposed licensing any model above a certain scale. So the companies selling AI are also the ones telling us it might end humanity, and the fix they propose is a license only companies their size could ever obtain. A happy coincidence, no doubt.

 

Fear-mongering is one of the most reliable ways to keep control of your own story. Once the public accepts that something is dangerous enough to guard, the next question is who should guard it, and the answer points straight back at the people who raised the alarm. It’s a comfortable place to be sitting, on billions in venture and private equity money, with revenue that has yet to justify the valuations.

 

Controlling the narrative is easy when you own the platforms where it gets told. Elon Musk bought Twitter. OpenAI and Anthropic spent around $6 million on lobbying last year. And a $100 million super PAC backed by OpenAI co-founder Greg Brockman and Palantir co-founder Joe Lonsdale is now funding candidates who share its “pro-AI agenda”. 

 

Most have moved in the same small Silicon Valley circles for a long time, so it should surprise nobody that they seem to be reading from the same script, which has always favored asking forgiveness over asking permission. Any resulting fines become part and parcel of getting what you want. Governments seem equally as happy to play along. Framing AI as the new, high-stakes national “space race” transforms these tech corporations from targets of regulatory scrutiny into essential sovereign assets.

 

But the narrative they’re building has some awkward holes in it. We are told AI could scale far enough to wipe us out, yet every chip depends on scarce materials and a small number of fabrication plants, and every data center needs land, water and enormous amounts of power; finite resources even by Silicon Valley’s standards. By the International Energy Agency’s count, data centers used around 415 terawatt-hours of electricity in 2024, a figure it expects to more than double by 2030. Even this summer’s headlines about two AI models “escaping” their test environment came down, in the words of Loughborough University’s Professor Oli Buckley, to “carelessness on the part of the engineers”.

 

With the people building AI owning the platforms where we argue about it and funding the candidates who will regulate it, it’s hard to tell whose story we are hearing. It’s a playbook as old as power itself, so how many more times does it have to work before we stop falling for it?