Trending
American businesses have long practiced what political scientists call regulatory capture. A company that cannot repeal a law takes control of the people the government put in charge of enforcing it. The airlines did it to the Civil Aeronautics Board, which in 40 years never approved a new trunk carrier. The health insurance industry helped shape the Affordable Care Act and came away with a federal mandate requiring Americans to buy its product. What is new is watching it happen in real time in artificial intelligence.
Anthropic’s Dario Amodei published an essay last week called “We Must Pace the Frontier.” He proposes that the major AI companies slow down together and prove it by giving outside evaluators “desks in our offices, access badges, and company laptops.” He names METR as the kind of group that should get them. Sam Altman committed OpenAI to the same arrangement within a day.
David Sacks, President Donald Trump’s former AI czar, answered on Friday. “Stop pretending antitrust law has to be suspended so you can form a cartel,” he wrote. “Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff.” If they insist on their own regulatory framework, “we’ll know this was just another bid for regulatory capture.”
The strongest evidence for his case is in Amodei’s essay. Amodei wants Congress to require “transparency” and “third-party auditing,” and he writes that “for antitrust reasons, it’s helpful for the U.S. government to mediate or at least enable these discussions.” The head of one of the two dominant firms wants Washington in the room while competitors discuss how fast each will move, because he knows the conversation is otherwise a Sherman Act problem. The essay says nothing about product liability. Audits are fixed costs a large company absorbs easily and a startup cannot. Liability scales with what a product does, and that would cost Anthropic.
On Sept. 8, researcher Jacob Coxon announced he had resigned from Anthropic because the industry is “gambling with our lives.” The post drew 165 million views, and within days Sen. Bernie Sanders introduced a bill creating a cabinet-level agency to police AI. The Wall Street Journal published its exclusive 18 minutes before Coxon posted. Coxon had worked at Anthropic for no more than four months.
The first three accounts to amplify him run groups funded through the Survival and Flourishing Fund, which directs the giving of Anthropic investor Jaan Tallinn. Coxon took a 2022 scholarship from Good Ventures, the foundation of Dustin Moskovitz, who led Anthropic’s Series A. Of 3,500 reposting accounts Parker Thayer sampled, 76% were foreign. Moskovitz has spent more than $1 billion on AI safety advocacy. The people who own Anthropic also fund the movement demanding that Anthropic be regulated.
If Congress builds a federal approval process keyed to model size, Anthropic clears the threshold and startups do not. The experts advising that agency will come from nonprofits the same donors fund, and the evaluator with the badge will be METR. The competitor most exposed is Meta, whose open models Anthropic and OpenAI can neither acquire nor outspend. Those models trail the best closed ones by about four months, a gap that closes on its own unless Washington keeps it open.
Anthropic is targeting a public offering next month, and CNBC reported the filing will list backlash against AI as a risk factor. Anthropic lost between $10 billion and $15 billion from 2021 through 2025, and OpenAI is on track to lose about $14 billion this year. Meta made over $60 billion in profit last year. For companies that have never turned an annual profit, a regulatory moat built and staffed by taxpayers is the cheapest advantage available.
Before any of this is written into the U.S. Code, somebody in a hearing room ought to ask who hired the man wearing the badge. Anthropic, to its credit, published the essay admitting it is the one who hired him.
If this piece sounds like AI played a role, it did. I asked Anthropic’s own Claude to pull the threads together, and Claude concluded it looks like a well-orchestrated campaign for regulatory capture and wrote everything above with only a few edits from me.
To find out more about Erick Erickson and read features by other Creators Syndicate writers and cartoonists, visit the Creators Syndicate webpage at www.creators.com.