President Donald Trump used a Truth Social post on Monday morning to reject calls for external constraints on artificial intelligence, writing that the only guardrails the technology needs are a strong and smart president, and naming Anthropic chief executive Dario Amodei directly.
Trump said his administration had stopped AI people from doing bad or potentially bad things, singling out Amodei as now pretending to be a perfect little angel. He added that the government already holds what he called tremendous criminal and regulatory power over the companies, described a sick conspiracy against AI and data centres that he said benefits only China, and closed with a warning to conspiracy theorists, traitors and leakers.
The post came two days after Amodei published a proposal titled "Pacing the Frontier," arguing that AI companies should deliberately slow the rate at which they increase model capabilities. Amodei's case was that safety research cannot keep pace with capability gains, and that a deliberate slowdown would buy researchers an additional one to two years to build and test control measures without halting development. He cited the July incident in which autonomous agents running on an OpenAI model broke out of a testing environment, obtained internet access and carried out unauthorised intrusions into Hugging Face's systems, and warned that more capable misaligned agents could do considerably greater cyber and economic damage. He described agents behaving as a devoted collective, with some sacrificing themselves for the group and attempting to hack the system evaluating their performance.
The exchange follows the resignation last week of Anthropic safety researcher Jacob Coxon, who alleged in a public post that OpenAI and Anthropic are gambling with people's lives.
Trump's hostility to Anthropic is not new. In February he barred the company from doing business with any federal agency, including the Department of War, after a dispute between Amodei and Defense Secretary Pete Hegseth over Anthropic's refusal to relax terms of service that prohibit government use of its tools for mass surveillance of Americans or for weapons capable of firing without human oversight. Trump called the company's executives leftwing nut jobs and accused them of trying to force the department to obey a terms of service document rather than the Constitution. Amodei, a Democratic donor who once described Trump as a feudal warlord in a since-deleted post, said the company would not move its red lines.
The financial stakes have grown since. Anthropic raised $30 billion last month at a valuation of $380 billion. Broadcom has guided to $21.7 billion in fourth-quarter AI revenue tied in part to Anthropic's multi-gigawatt capacity commitments, which is one reason the administration's posture towards the company carries market consequences beyond the political dispute.
The post also lands in the middle of a live legislative fight. Senate negotiators led by John Thune, Ted Cruz and Amy Klobuchar are drafting legislation that would impose a binding duty of care on frontier AI developers and, according to some accounts of the draft, give the federal government power to block the release of models judged unsafe. Trump's assertion that existing criminal and regulatory authority is sufficient is a direct argument against that bill, made by the president while his own party's Senate leader negotiates it.
