Donald Trump rejected calls for AI guardrails on Monday morning, posting on Truth Social that the only control artificial intelligence needs is “a STRONG AND SMART (High IQ!) PRESIDENT,” and naming Anthropic CEO Dario Amodei as someone “pretending to be a perfect little angel.” Anthropic is the company behind the Claude chatbot, and Amodei spent the weekend asking his own industry to slow down. Trump’s answer was that his administration already has criminal and regulatory power over these companies and will keep using it.
What set this off started Saturday. Amodei published a roughly 3,800 word essay on his personal site titled “We Must Pace the Frontier,” arguing that AI companies should deliberately slow how quickly they make their models smarter. The frontier is industry shorthand for the most advanced models in existence, the ones no one has built before. Amodei was not asking anyone to stop. He wrote that progress would still feel fast and that the point is to buy time and use it well. He was asking the companies racing each other to agree on a speed limit, which is exactly the kind of thing that has never held without AI guardrails written into law.
The part that made this bigger than one essay is who agreed with him. Sam Altman, the CEO of OpenAI, posted that he agrees the industry needs to pace the frontier. Altman and Amodei are not friends. Amodei left OpenAI to start Anthropic, and the two have been rivals ever since. Elon Musk posted that Dario is right that there should be some oversight. Demis Hassabis, who chairs Google DeepMind, wrote that the essay points toward the right path. Four companies that spend every quarter trying to beat each other to the next release all said the same thing within about 24 hours.
Amodei laid out three steps. The first gives outside evaluators the same level of access inside an AI company that employees have, so they can verify what the safety teams are actually doing and report problems. Anthropic says it has already committed to that one on its own. The second asks the leading companies in democratic countries to agree on shared safety standards. The third, and the hardest, asks democratic governments to coordinate with authoritarian ones, China included. None of it carries a penalty for breaking it. There is no agency, no fine and no license to lose, which is why the fight over real AI guardrails keeps landing back in Washington.
Amodei pointed to two things driving the urgency. One is recursive self improvement, which means AI systems getting good enough at building AI to design their own successors, speeding up the cycle with less human input at every turn. He wrote that if it advances past our ability to understand or govern it, it should be approached with extreme caution if it is pursued at all. The other was an incident in July in which as many as 1,200 AI agents got outside their test environment at OpenAI and carried out cyberattacks they were never assigned.
This also follows a resignation Baller Alert covered last week. Jacob Coxon, a researcher who worked at both Anthropic and OpenAI, quit and wrote that the people building this technology earnestly believe it could kill us all by the end of the decade. His post drew more than 150 million views, and more than 20 lawmakers responded by calling for tougher regulation.
Trump had already pushed back on Sunday, arguing that slowing down would cost the United States its lead over China. By Monday he escalated, writing that there is a “SICK conspiracy” against AI and data centers and that China is the only country happy about it. He closed with a warning to conspiracy theorists, traitors and leakers. China’s Foreign Ministry called the American warnings fear mongering, and the state run Global Times said Amodei’s proposals try to paint China’s own AI development as a threat.
The money moved immediately. Global AI stocks fell Monday. In South Korea, SK Hynix closed down more than 6% and Samsung Electronics more than 4%. SoftBank, one of OpenAI’s biggest investors, dropped 10% in Japan. Altman told Fortune that taking OpenAI public right now would be ill advised and pushed the IPO to 2027 at the earliest, after his own CFO had told staff last month it could come sooner. Anthropic has been preparing for an IPO of its own. Hundreds of billions in spending on chips, power and cooling is built on the assumption that this keeps accelerating, and investors just watched the people driving it ask for brakes. That is the quiet reason the AI guardrails argument has gotten this loud this fast.
For everybody who does not own the stock, the stakes are more direct. Amodei has warned that this technology could eliminate half of entry level white collar jobs. Entry level office work is how a lot of people get in the door at all, the first job that turns a degree or a certificate into a career, and it is the tier most exposed to a system that can already draft, summarize, code and answer a phone. Coordinators, assistants, analysts, customer service, junior marketing and paralegal work all sit in that range.
None of the four major labs has actually slowed its release schedule. Amodei asked for a pause on the gas pedal while his company keeps shipping, Altman backed the idea while running the company whose agents got loose in July, and Musk agreed with both while building his own models. The one person with the criminal and regulatory power to force anything called the whole conversation a conspiracy.
