{"slug": "theres-a-right-way-to-regulate-ai-and-a-very-wrong-way", "title": "There’s a Right Way to Regulate AI—and a Very Wrong Way", "summary": "President Donald Trump rejected calls from Anthropic CEO Dario Amodei, OpenAI's Sam Altman, and xAI's Elon Musk to regulate artificial intelligence, calling the danger of AI \"a HOAX\" in a Truth Social post. The three rival AI chief executives had agreed that AI has become too dangerous not to regulate, while venture capitalist David Sacks and Cohere CEO Aiden Gomez accused Anthropic of running a \"sophisticated regulatory capture strategy based on fear-mongering.\" The dispute centers on whether federal AI regulation would improve safety or entrench the dominant AI companies' market power.", "body_md": "You’ve probably heard the one about the masochist who asks the sadist to beat him up. The sadist says no. That hoary joke came to mind when I saw President Donald Trump’s reply to Anthropic chief executive Dario Amodei’s [statement](https://darioamodei.com/post/we-must-pace-the-frontier) that artificial intelligence has become too dangerous not to regulate. Two rival AI chief executives, Open AI’s [Sam Altman](https://x.com/sama/status/2099348812305473766) and xAI’s [Elon Musk](https://x.com/elonmusk/status/2098789109980332057), agreed with Amodei. But Trump [said no](https://truthsocial.com/@realDonaldTrump/posts/117270591511950591):\n\n[W]hen, in the History of Business, did anyone see the Leaders of an Industry call for Regulation that, if strongly implemented, will drive them into oblivion and bankruptcy? AI taking over the World, destroying Humanity, and all other things bad, is a HOAX, no different from RUSSIA, RUSSIA, RUSSIA — UKRAINE, UKRAINE, UKRAINE — IMPEACHMENT HOAX #1 — IMPEACHMENT HOAX #2 — and all of the other HOAXES and SCAMS that America was forced to endure through the Destructionists’ and Deviants’ foul play and illegal conduct.\n\nI bet you feel better already!\n\nFrom a strictly medical point of view, my analogy of the masochist and the sadist does not apply. Amodei, Altman, and Musk can be faulted for not banding together earlier to demand forcefully that AI be regulated at the federal level. (Didn’t they see [*The Matrix*?](https://www.amazon.com/Matrix-Keanu-Reeves/dp/B001XVD2Z0/ref=tmm_aiv_swatch_0)) But procrastination isn’t in the Diagnostic and Statistical Manual of Mental Disorders (though being a billionaire perhaps should). For his part, Trump isn’t a sadist, exactly, but rather suffers from the related disorder of [malignant narcissism](https://www.theguardian.com/us-news/2024/oct/24/trump-nyt-ad-george-conway-pac), worsened lately by a [vanishing attention span](https://www.yahoo.com/news/politics/articles/80-years-old-trump-become-090000375.html). It’s our hard luck that a new global crisis is arising under the most mentally unstable president ever to occupy the White House—worse even than the heavy-drinking, deeply paranoid Richard Nixon, at the height of the Watergate scandal, [talking to portraits](https://www.amazon.com/Final-Days-Bob-Woodward/dp/0743274067) on the White House walls. The example of how Trump handled his last global crisis—the Covid epidemic—is [deeply discouraging](https://pmc.ncbi.nlm.nih.gov/articles/PMC9115435/). \n\nStill, Trump isn’t entirely wrong to be suspicious. The tech industry [has shown itself to be untrustworthy](https://newrepublic.com/article/208876/tech-world-evil-musk-bezos-thiel) in recent years, and some degree of skepticism about how Amodei, Altman, and Musk would like this to go forward is warranted.\n\nThe main suspicion is that these AI chief executives see regulation as an opportunity to reduce competition and otherwise benefit their own dominant companies, a process known as [regulatory capture](https://bfi.uchicago.edu/wp-content/uploads/2023/02/3003160.pdf). The tech venture capitalist David Sacks, who co-chairs Trump’s Council of Advisors on Science and Technology, [said on X](https://x.com/EvanHub/status/2097497037956891126) that “Anthropic is running a sophisticated regulatory capture strategy based on fear-mongering.” Aiden Gomez, chief executive of a Canadian AI startup called Cohere, [wrote](https://cohere.com/blog/who-gets-to-define-the-rules-for-ai) on his company website: “These oligopolies are now requesting to bend competition rules and be permitted to dictate the terms for everyone else. A wolf in sheep’s clothing, a cartel by any other name.” \n\nGomez is right to question some of what Amodei proposed in last week’s memo. The Anthropic chief’s first idea was that every “frontier AI company” commit to giving “employee-like access” to “a team of embedded third-party evaluators” who will impose safety standards. AI is “unilaterally committing to this step now,” Amodei wrote, and that’s laudable. But Amodei undermined his argument when he analogized these evaluators to bank regulators. The federal agencies that regulate banks—the Office of the Comptroller of the Currency, the Federal Deposit Insurance Corporation, even the Federal Reserve—are pretty much [the poster child](https://www.gao.gov/products/gao-20-519) for [regulatory capture](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=892925). Bankers are expert at persuading regulators that this or that regulatory move will bring down the whole house of cards. They’re able to bluff this way because finance, much like AI research, is a fast-changing realm that almost nobody really understands.\n\nAmodei’s second proposal was that “frontier AI companies within democratic countries coordinate to establish common safety standards as well as limits on the rate of unchecked AI progress.” This “will require government support,” Amodei wrote. That’s a nice way of saying it’s illegal under antitrust law. These “frontier AI companies within democratic countries” are Open AI, Anthropic, and perhaps one or two others, and if they consulted with each other as Amodei proposed they’d be operating like a cartel. Gomez is right to use that word.\n\nAmodei’s third proposal is unassailable. It’s international cooperation to guarantee safety worldwide. To Trump that sounds like surrendering to China. But it is by far the most necessary step, and nobody with a properly functioning cerebral cortex can oppose it.\n\nTo acknowledge that Amodei’s first two proposals would give Anthropic an unfair competitive advantage needn’t be to argue against regulation. It can simply be to argue against faulty regulation. Amodei’s first proposal might be made to work if the third-party evaluator were less like the usual run of bank examiners and more like an official at the Consumer Financial Protection Bureau, or CFPB. Trump put the CFPB on ice precisely because banks (and their allies in Congress) resented its focus on the common good rather than on banking profits. A third-party evaluator who acted more like a CFPB regulator would be helpful. A third-party evaluator who acted more like an FDIC regulator probably would not.\n\nAmodei’s second proposal can’t be made to work, but it’s a useful illustration of why AI needs to be regulated by government. The only way to set proper industry standards is for a bunch of experts to get together and hammer them out. But having the individual companies do this on their own is unacceptable because it’s anticompetitive. So instead the job typically gets done by government regulators in consultation with the companies, with public interest groups, with concerned academic experts, and various other “stakeholders” (a word I tend to avoid because it makes me think of a large cut of sirloin dangling from a grill fork). If there’s no statute under which the government may plausibly do this, then Congress should create one after conducting hearings and holding meetings that address the more obvious aspects of the problem, while leaving abstruse details to regulators.\n\nWhatever the process, it needs to begin yesterday, because [recursive self-improvement](https://www.cnbc.com/2026/09/11/anthropic-openai-ai-existential-concerns.html)—the ability of AI to create the next generation of AI without human intervention—is right around the corner. Three weeks ago, how many of us had even heard of recursive self-improvement? Now it’s scaring us to death.\n\nOpposition to regulating AI may not be so formidable as it seems. Venture capitalists like Sacks and Marc Andreesen are the most fervently opposed because they’re all about money. But tech entrepreneurs (omitting the [Antichrist-obsessed](https://www.wsj.com/tech/peter-thiel-antichrist-lectures-dd28c876?eafs_enabled=false) Peter Thiel) are lately showing more sympathy toward the idea, perhaps because they’re more familiar with what AI can do. Gomez, for instance, after denouncing Amodei’s regulatory scheme, proposed one of his own, including an “independent testing” scheme that sounded a lot like Amodei’s bank-examiner proposal, only not with a risk framework defined by “a select few companies.” (Gomez even repeated Amodei’s clueless suggestion that bank regulators might provide a sound model!) Gomez’s definition of what constituted risk was too complex and technical for me to understand, but if the gist is that the focus should be on the really dangerous stuff, count me in.\n\nTrump probably never heard of regulatory capture. But he’s right to observe that it’s unusual for a company to judge *itself* a threat, as Anthropic did last week. Jacob Coxon, a 27 year-old AI researcher at Anthropic, quit his job and [posted on X](https://x.com/EvanHub/status/2097497037956891126) a tweet that quickly circled the globe: “The people building AI earnestly believe that it could kill us all by the end of the decade.” In response, [a top Anthropic scientist replied](https://x.com/EvanHub/status/2097497037956891126): “Jacob is correct here—we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to.” With Amodei posting his memo shortly thereafter, Anthropic was fairly shouting: Stop me before I kill again!\n\nIf the Frankenstein monster came up to you and said, “Stop me before I kill again,” you wouldn’t ask it which safety precautions to take. You’d lock it up. The fear among AI leaders is persuasive. If they’re afraid, then I’m going to be afraid, too. But finding the solution to ameliorate these fears shouldn’t be left to them. Our democratic institutions of governance are built to handle this. Electing a Democratic Congress in November will help speed the process along. Let’s hope Trump can be persuaded to get out of the way.", "url": "https://wpnews.pro/news/theres-a-right-way-to-regulate-ai-and-a-very-wrong-way", "canonical_source": "https://newrepublic.com/article/215472/trump-regulate-artificial-intelligence-capture", "published_at": "2026-09-16 10:00:00+00:00", "updated_at": "2026-09-16 11:42:45.358123+00:00", "lang": "en", "topics": ["ai-policy", "ai-safety", "ai-ethics", "artificial-intelligence"], "entities": ["Donald Trump", "Dario Amodei", "Anthropic", "Sam Altman", "OpenAI", "Elon Musk", "xAI", "David Sacks"], "alternates": {"html": "https://wpnews.pro/news/theres-a-right-way-to-regulate-ai-and-a-very-wrong-way", "markdown": "https://wpnews.pro/news/theres-a-right-way-to-regulate-ai-and-a-very-wrong-way.md", "text": "https://wpnews.pro/news/theres-a-right-way-to-regulate-ai-and-a-very-wrong-way.txt", "jsonld": "https://wpnews.pro/news/theres-a-right-way-to-regulate-ai-and-a-very-wrong-way.jsonld"}}