Dario Amodei is Silly Wonka
Anthropic’s CEO thinks we should slow him down. His competitors agree. The American people certainly agree. Unfortunately Donald Trump vehemently disagrees, saying all it takes to govern AI properly is a “smart president.” Paraphrasing Indigo Montoya, “I do not think that means what you think it does.” Providing an even bigger wrinkle is China. China also disagrees. Now we’re in quite the pickle.
Geopolitics is ultimately the answer here, but let’s indulge these AI tech mavens for a moment.
They’re all terrible. Let’s start there. Elon’s a nazi. Sam is a sociopath. Even the Google twins have taken a MAGA turn. But the smartest one of the bunch is passing himself off as the one who’s really looking out for us. Our AI savior. In a 3,800 word letter, Anthropic founder Dario Amodei warned the world yet again of the dangers of AI. That we must work to slow the progress of development lest we be vaporized as a species, or something to that effect.
His claim, that he repeated yet again on CBS Sunday Morning, is that private industry cannot be left unchecked, the frontier models are already breaking bad, the government should have oversight or even be involved in these companies, we need an updated biological weapons agreement with China and htat we should also stop selling chips to them. You know, just in case.
Let me break this down another way using his sentiments, because the framing around Amodei is that he’s the responsible one in the room. Even people like Elon Musk have said that publicly. That’s why he left OpenAI in the first place—Altman was too much of a wild card and wasn’t building responsibly. That’s the Anthropic origin story and it has been a winning narrative.
I’m over it. I’ve been over it for a long time.
Methinks he doth protest too much. He’s suggesting that he can be trusted since he’s the one advocating for some nebulous regulations and oversight. For some magical body governed by a collection of nations with democratic elections to set the standards that presumably he’s establishing over at Anthropic. But he already has models that are breaking out and doing unruly things. So whose standard are we working towards?
Which nations qualify as democratic? Is Ecuador invited to this party? Argentina with nutty Milei at the helm? Germany has free and fair elections and one of their states just elected the lineal heir to the nazi party. How about Turkey? Their elections are pretty good. Not the best, but what’s the benchmark? And how would these regulations be governed? Like NATO? That’s been working well. Like multilateral trade agreements? We’ve personally blown those to shit.
Who exactly is Dario appealing to? Congress? The American people? Donald Trump? It must be Trump, right? After all, he declared that the United States was going to be the crypto capital of the world and first in AI. And he did the bidding of the AI companies by prohibiting the sale of the most modern chips to China, something Dario said was a good idea and that we should stick to. And that’s because when pressed on what happens if China suddenly leapfrogs the United States in AI technology that would be a disaster because then China would have better and perhaps more reckless AI that could ruin the world.
But they’re not democratically elected. So would they be invited to Dario’s party?
But let’s stay on Trump before digging further into the arms race with China. All of Silicon Valley got behind Trump because it was concerned that the Democratic administrations would slow the development of this technology and put guardrails in place. This administration campaigned on literally destroying the regulatory state entirely so that corporations are free to do as they please. To be clear, Dario Amodei didn’t donate to Trump nor does he support this administration. But if that’s the case, then say the quiet part out loud. Be an actual hero. If you want the government to slow the development of AI then you need a different government. You’re shouting at the rain asking for Trump to slow this down.
And since he’s in control of both houses of Congress I can only assume that when Dario cries out for help like this he’s talking to the American people. Is he asking us to throw out the Republicans at the midterms and replace them in 2028 as well? Because if so, that would be a courageous ask. But he didn’t do that. Won’t do that.
Instead, he’s stoking fear about the thing that he is not slowing down on himself. Why? Because this is capitalism. This is a freight train and he’s been in the engine room of this thing since the beginning. Shouting over the noise of the engine and the running train at some signal operator with headphones on a few miles away to change tracks is hardly an effective way to get this done.
This isn’t courageous and we need to stop acting like it is. Courage is Daniel Ellsberg putting his life and job on the line to get the Pentagon Papers out in public. It’s Chelsea Manning sending evidence of war crimes to Wikileaks. It’s Gary Webb publishing the DC connections to the crack epidemic in our cities. It’s Thomas Drake and John Kiriakou exposing intelligence agency corruption. Courage would be shutting the entire company down, getting sued by your investors, being publicly shunned.
That’s not what he’s doing. Dario Amodei is Willy Wonka. He’s whispering “no, stop, don’t” and letting the kids harm themselves anyway. So he can say, “I tried to warn them” even as they come out the other side paper thin, blueberry colored or covered in trash.
In business terms he’s building what I call “a tobacco moat.” Dario isn’t interested in regulations. If he was he would have suggested what they actually look like and advocate for a complete government takeover of the companies. When asked about whether he would hand over Anthropic to the government he said, partially and “to the right combination of governments.” And we’re right back to where we started. Which ones? How would that look? Who’s in charge? Blah, blah, blah. It’s a feint. A misdirect.
What he’s looking for is a regulatory maneuver that would prevent new entrants into the market. It would allow the early adopters to maintain the pole position. Much in the way the tobacco giants welcomed the abolition of advertising because it meant no startup would ever be able to gain market share, Dario Amodei is looking to cement his lead now that his customers are addicted. Customers, by the way, that still include the United States military despite the very public brouhaha between him and Hegseth.
Amodei claims that AI started in private industry and that government has to step in now, after the fact, to cool it down. That’s like claiming the war in Gaza started on October 7th. Like every other innovation since the industrial revolution, AI started in government agencies and university laboratories that were funded by government agencies. AI started with a foundation research grant to a university in the 1950s. It went from the lab to the government and was developed by DARPA around the time it also created the internet. Every one of these so-called AI geniuses is operating with technology and concepts that were developed by the government and/or government grants. It wasn’t until they fucking broke loose and tried to capitalize on it that we found ourselves here.
I’m reiterating this longstanding point about the fallacy of private industry innovation because he doesn’t just get to pick a start date, build one of the largest private investment vehicles in the world around claims like “we’re going to cure cancer” and “we can replace every workers so we’ll have paradise and leisure” and then say can you take a stake in us because we might just kill everyone.
So I ask again. Who are you talking to when you say “we” need to slow down the development of AI? If you’re the guy in the lead - something pretty much everyone acknowledges at this point - and every one of the investors that you courted are responsible for electing an administration that promised to let corporations do whatever the fuck they want… then who the fuck is “we” asshole?
Back to China. In the same interview where he suggests the nebulous government oversight, Amodei says that we must absolutely prevent China from acquiring our technology, mainly our chips. He calls this the “toughest dilemma” because if we slow down and they don’t then we lose. There’s only one place I think we should take him at his word. He says the negotiations should be viewed through the Cold War nuclear disarmament lens but that the talks are “above his paygrade.” He has thoughts, he says, but no one has asked him yet.
I have thoughts as well.
There seems to be consensus among AI whistleblowers that this idea that AI can destroy humanity within the decade is statistically legit. So I’m not here questioning the veracity of this claim. People said the same thing about nuclear weapons and here we are. We already know that humans are capable of realizing airborne pathogens that kill millions of people. Just lived through that, thanks. So the idea that “recursive self-improvement” of these models will allow AI the ability to launch those nuclear weapons or release a virus seems plausible.
The Lever’s David Sirota of Don’t Look Up fame—a movie that parodies ignoring the climate apocalypse with the device of a meteor—has suggested that this is a similar theme and that we could solve it by passing a law to jail the owners of the technology if it goes rogue and starts killing people. But I think there’s an easier parallel and old Dario basically made it. Stop the IPOs of these companies and nationalize the AI labs. Make the investors whole if you must but just take it all in house. And for the love of God, keep Elon out of the AI business.
And here’s why. As Dario suggests, we should take a Cold War framing on this issue. What do you think would have happened if we privatized nuclear weaponry? I can tell you part of that story. The Atomic Energy Commission decided after World War II that it would examine commercial uses for atomic energy by partnering with private industry and it was such a disaster they not only shut the program, they changed the name of the fucking agency. Now it’s the Department of Energy.
If private industry ran the nuclear program and had a profit incentive I’m sure it too would have had some executive in the 60’s being like, “uh, someone should really regulate this.”
If you run a technology that you think has the potential of ending humanity and that others might be able to do the same, this is government playbook shit. And we need serious people in government to take control of it, load it with bureaucratic failsafes, smother it in diplomatic entreaties to all other nations in a position to weaponize it and wrap it in ironclad international agreements. Otherwise, you’re just some smart dude with a conscience that is subservient to greed; aware enough to build a case for “I told you so” but not courageous enough to forego the riches that might come from it. Even if there’s no planet left on which you might enjoy them.
