You’ve seen the headlines. You’ve heard the warnings from the very people building the most dangerous technology in history. They tell you AI could destroy humanity. And then they say, ‘Trust us. We’ll handle it.’
Something stinks.
Over 1,100 employees of frontier AI labs signed a letter that basically says: Let us be the gatekeepers. We know better than you. They’re not just asking for safety standards. They’re asking for the keys to the kingdom. And they’re using existential fear to get them.
The people who are building the bomb are now telling you they’re the only ones qualified to guard the bomb.
This isn’t a safety movement. This is regulatory capture disguised as altruism. It’s a power play, and it’s working.
Think about it. The same companies that accelerated the AI race — pouring billions into training ever-larger models, ignoring red flags, cutting corners on alignment research — are now positioning themselves as the responsible adults. They created the crisis. Now they want to be the only ones allowed to solve it.
That’s not a contradiction. That’s a strategy.
By pushing for strict licensing, export controls, and compute governance, they create a moat that no startup, no open-source project, no rival can cross. ‘Safety’ becomes a barrier to entry. The ladder gets pulled up.
Safety isn’t a principle. It’s a moat.
You’ve probably felt this paternalistic tone before. It’s the same attitude that gave us ‘move fast and break things’ — then, years later, a thousand apologies and a new set of walls. The tech elite doesn’t want to stop building AI. They want to stop you from building AI.
I saw this firsthand at a recent conference. A senior researcher from a frontier lab gave a talk on ‘AI Safety.’ The first slide was a list of reasons why only labs with $100 million+ compute budgets should be allowed to train frontier models. No mention of the irony. No one laughed.
This is dangerous. Not because AI is dangerous — though it might be. But because the cure they’re proposing is worse than the disease. A small group of unelected, profit-driven actors get to decide who builds the future. They get to set the rules. They get to enforce them. And they get to profit from the scarcity they create.
Existential risk is the most effective anti-competitive tool ever invented.
So what do we do? First, stop believing the hype. The same people who told you AGI was coming in 2 years are now telling you only they can save us. They can’t have it both ways.
Second, demand transparency. If safety is the goal, the regulations should apply to everyone equally — including the incumbents. No special exemptions. No grandfather clauses.
Third, support open-source alternatives. The only real antidote to gatekeeping is a decentralized, transparent, and permissionless ecosystem. The more people who can build and audit AI, the harder it is for a few to lock it down.
The future of intelligence should not be owned by a handful of Californian labs. The future of intelligence should be ours.
Don’t let them scare you into giving up your power.
FAQ
Q: Isn't it reasonable to want safety experts to have control over AI development?
A: It would be if the 'safety experts' weren't also the people racing to build the most powerful systems. The conflict of interest is glaring: they create the risks, then claim only they can manage them. Real safety requires independent oversight, not self-regulation by the same labs.
Q: What does this mean for open-source AI projects?
A: If the incumbents succeed in framing safety as requiring strict licensing and compute governance, open-source models will be regulated out of existence. This kills innovation, transparency, and the democratization of AI. The real goal is to eliminate competition, not to protect the public.
Q: But what if the frontier labs are right about existential risk?
A: Even if existential risk is real, handing sole control to the labs that created it is the worst possible solution. It creates a single point of failure and a massive incentive to downplay dangers. We need diverse, democratic oversight — not a self-appointed priesthood of engineers.