AI Doom Is a Sales Pitch, Says a Former FTC Commissioner
Alvaro Bedoya, who sat on the Federal Trade Commission until his disputed 2025 removal, is pushing back on the existential AI warnings coming from the biggest labs. His case rests on a simple observation about who benefits from the fear, and the enforcement record mostly backs him up.
Is AI going to kill everyone? Alvaro Bedoya doesn't think so, and the former US Federal Trade Commission commissioner isn't hedging about it.
Bedoya served on the commission from May 2022 until his disputed removal in 2025. His post this week pushes back hard on the existential doom warnings coming out of the biggest AI labs. The argument, roughly: humanity has survived worse, and the loudest alarm bells are ringing from the people selling the thing they're warning you about. The question worth asking: who profits from telling you a machine might end the world?
The Numbers Behind the Noise
Start with who's doing the warning. OpenAI raised money at a $157 billion valuation in October 2024. Anthropic was valued at $61.5 billion in March 2025. Those are the labs most vocal about catastrophic risk. Granted, size alone doesn't make an argument wrong. But it does mean the doom thesis and the business model sit in the same building.
Now look at what the FTC actually went after. In September 2024, the agency announced Operation AI Comply, a sweep that produced five enforcement actions in a single day. The targets weren't sentient machines. They were companies allegedly selling AI legal advice that wasn't legal advice, and AI-written reviews that were simply invented. Real harms. Mundane harms.
The FTC also logged $12.5 billion in consumer fraud losses for 2024. That's the ground Bedoya spent his tenure on, from data brokers to shady surveillance to scams that use new tools the old way.
History Has a Track Record Here
Every generation gets its apocalypse. Y2K was going to crash the grid. Nuclear war was going to end civilization by 1960. The printing press was going to wreck the church and the state at once. "Humanity has survived worse" is a cliché because it keeps happening.
I'm not sure past panics are a reliable guide to this one. But they're not nothing, either. The record shows that doomsayers get the direction roughly right and the timeline wildly wrong, then quietly retire the prediction when the date passes with nobody noticing.
Where the Insiders Land
According to Bedoya, the doom framing does real work inside the industry. It justifies keeping models closed, it invites regulators to trust incumbents with oversight, and it turns a business question into a civilizational one. Skeptics of that framing have been making this point for two years now, and the argument keeps gaining ground.
Color me skeptical, but the "this is too dangerous for anyone but us to build" pitch has always been a tidy way to keep competitors out. Proponents of strict licensing say the stakes justify the caution. That's a defensible position. It's also a convenient one.
What to Watch
Three things. First, the FTC's current enforcement pattern on AI claims, which will tell you more about real regulatory risk than any white paper. Second, the state laws now hitting the books. Colorado's AI Act was pushed to June 2026, California's SB 53 is live, and the EU's obligations for general-purpose AI models kicked in during August 2025. Third, watch whether Congress moves to preempt state rules. That would reset the whole map.
Bedoya's own fight isn't finished. A federal judge ruled his removal unlawful, and the case is still winding through the courts. If he's right about the doom talk, those rulings will matter more than the labs' warnings ever did.
His bet is that the boring harms win the argument in the end. On the evidence so far, that's the better bet.
