You’ve felt it. That nagging suspicion every time a Silicon Valley billionaire gets on a stage and begs the government to regulate them. Dario Amodei, CEO of Anthropic, just published a manifesto titled “We Must Pace the Frontier.” It sounds noble. It feels safe. But it is a lie by omission.
When the loudest voices begging for safety regulations are the ones holding the keys to the castle, you aren’t watching a rescue mission. You’re watching a drawbridge being pulled up.
Amodei wants us to believe that slowing down AI development is about protecting humanity from hypothetical existential risks. But read between the lines of his polished corporate prose. As one frustrated commenter pointed out immediately, “This certainly looks like a way to slow down competitors and regulate foreign and open models. It’s always about money.”
Here is the twist you probably missed: The people pushing back against Amodei’s vision aren’t a bunch of reckless tech-bros who want to build Skynet. They are developers and citizens who are fundamentally anti-incumbent centralization. They don’t distrust safety; they distrust unaccountable power resting in the hands of three Western tech giants.
The same policy that supposedly protects humanity from hypothetical frontier risks also protects Anthropic from real, immediate market threats. “Pacing the frontier” sounds like control for safety, but it reads like control of the market.
You cannot separate the sincere desire to save humanity from the strategic need to starve your competitors. In Silicon Valley, altruism is just monopoly wearing a hard hat.
Look at the actual users in the trenches. One developer who pays for Anthropic’s premium Claude Code wrote: “I enjoy Claude Code very much… but the doomer marketing and this whole regulate-while-we’re-ahead spiel is extremely annoying and makes me wish Anthropic gets trounced.” The people actually building with AI are calling BS. They see the “pacing” argument for what it is: an attempt to lock the open-source ecosystem out of the market.
Another commenter proposed a simple alternative: “Anthropic releases models as open weights + more information about how they do training and alignment.” But Anthropic won’t do that. Why? Because giving away the keys breaks the fortress.
This isn’t a debate about whether AI is dangerous. It’s a debate about who gets to own the future. If Amodei gets his way, AI doesn’t become a public utility. It becomes a heavily regulated, government-sanctioned fortress controlled entirely by a handful of elite labs. The outcome of this debate decides whether AI remains an open tool for developers worldwide, or becomes a gated community for the highest bidder.
Safety is the perfect marketing gimmick because it’s the one argument that demands you stop asking questions.
The next time a billionaire tech CEO asks the government to slow them down, ask yourself who they are actually trying to slow down. The frontier isn’t being paced. It’s being privatized.
FAQ
Q: Is Dario Amodei actually faking his concern for AI safety?
A: It's not about faking it; it's about inseparability. His sincere fear of AI risks and his corporate incentive to block competitors are pointing in the exact same direction. You can't separate the good intentions from the self-serving strategy.
Q: How does 'pacing the frontier' actually hurt open-source developers?
A: If governments mandate massive compute audits, safety tests, and licensing fees for 'frontier models,' only multi-billion-dollar labs can afford to compete. It turns AI development from an open ecosystem into a gated country club.
Q: If open-source wins, doesn't that mean anyone can build dangerous AI?
A: Yes, that's the real risk. But the anti-incumbent crowd argues that decentralized power is safer than concentrating the most powerful technology in human history in the hands of three unaccountable CEOs.