President Trump unleashed a barrage of social media posts on Monday, sharply criticizing industry leaders who are urging a slowdown in the pace of AI model development. His statements come in direct response to a weekend column by Anthropic CEO Dario Amodei, which called on governments and the broader industry to adopt more comprehensive AI safety standards. Experts suggest that without presidential backing, and given the significant divisions among lawmakers on the issue, the prospect of federal AI regulation in the United States in the near term remains highly unlikely.

On April 30, 2026, in San Francisco, California, Dario Amodei, co-founder and CEO of Anthropic, gave an interview to The Circuit with Emily Chang at the company's headquarters. Over the weekend, executives from Anthropic, OpenAI, and other leading AI firms publicly voiced warnings about the potential dangers of advanced artificial intelligence and urged a more cautious approach to model development. Trump's Monday posts, however, dismissed these appeals outright. In one post targeting tech executives and their critics, he wrote: "AI doesn't need regulation or 'guardrails' - it just needs a strong, smart (High IQ!) President, and the United States has exactly that, and plenty of it!"

Trump's open opposition to new AI rules presents a fresh dilemma for Anthropic and OpenAI. These companies must sustain rapid growth to justify their near-trillion-dollar valuations, while concurrently addressing growing societal concerns that AI's rapid advancement could slip beyond human control. The recent departures of two safety researchers from Anthropic, Jacob Cawson and Joe Benton, highlight deep-seated worries within the industry about potentially catastrophic consequences of AI systems. Against this backdrop, CEO Dario Amodei published his weekend piece calling for a deceleration in AI development, proposing an audit and oversight framework intended to avoid sacrificing commercial competitiveness or compromising U.S. national security.

OpenAI's Sam Altman and Elon Musk, in a rare show of public accord, expressed support for Amodei's stance. Amodei's proposal, targeted at an industry with minimal federal oversight, includes embedding third-party evaluators within frontier AI labs, establishing unified safety standards among democratic nations, and, where feasible, coordinating with governments in authoritarian regimes. He stated that Anthropic is willing to grant evaluators "near-employee level access" to monitor and verify model safety. He cited METR, a nonprofit founded by former OpenAI and DeepMind researchers, whose mission is to develop scientific methods for assessing catastrophic risks from AI autonomous capabilities to inform AI development decisions. Other organizations working in this space include FAR.AI, AVERI, and Apollo Research.

Tom Wheeler, former chairman of the Federal Communications Commission and visiting fellow at the Brookings Institution, expressed skepticism about relying on tech companies for meaningful self-regulation. "Who sets the standards? Who decides who performs the evaluation? Who is responsible for enforcement?" Wheeler questioned. While acknowledging that AI industry leaders recognizing "significant risks" is commendable, he characterized the latest industry oversight proposal as too vague, essentially a "hollow shell." Wheeler previously oversaw the media and telecommunications sector. In the U.S., nearly all major industries - from food and beverage, automotive, and pharmaceuticals to finance - are subject to strict federal oversight. Artificial intelligence, a key driver of the American economy, is a notable exception, with regulatory efforts left largely to individual states. California Governor Newsom, for instance, recently signed multiple AI bills, including the Adam Act, which mandates that AI chatbot providers protect minors from harmful content.

"Going to Extinction and Bankruptcy": Trump, who signed an executive order in June requiring only voluntary information sharing from AI companies before model release, now firmly opposes any federal oversight, calling concerns about AI risks "a hoax." He posted on Truth Social on Monday, "In the history of business, have you ever seen leaders of an industry call for regulation? If these regulations are strictly implemented, they will drive them to extinction and bankruptcy." Concurrently, David Sacks, Trump's former AI chief, continues to criticize Anthropic, accusing it of engaging in "regulatory capture." The venture capitalist, who has investments in several AI startups, responded to Amodei's column online, arguing that leading model companies should exercise self-restraint rather than petition the government. He stated, "If unreleased models are already dangerous enough to warrant a slowdown, I support companies making responsible choices on their own."

Representatives for Anthropic and OpenAI have not yet responded to requests for comment. The Trump administration's position is that any impediment to U.S. AI development benefits China, the world's second-largest economy and top geopolitical rival. This stance was on display at a major AI event in Shanghai, China in July. Jessica Ji, a senior analyst at Georgetown University's Center for Security and Emerging Technology, noted that the president is willing to use his power to punish companies that openly oppose his views. Anthropic has already experienced this pressure: earlier this year, it was affected by actions from the Department of Defense, and in June, government export controls forced it to shut off external access to two of its models. Ji commented, "These companies are very wary of being punished by the Trump administration. The authorities are effectively pushing companies to not pause research and development, but to continue iterating and releasing new models."

Beyond Amodei, Altman, and Musk, Microsoft CEO Satya Nadella voiced support on Sunday for a "prudent pace of AI development." Demis Hassabis, chair of Google DeepMind, stated on X that "Dario's article points in the right direction." Meanwhile, Meta CEO Mark Zuckerberg, who is rapidly iterating on his company's own models while attempting to align with the administration, has not yet made a public statement and Meta did not respond to interview requests.

One practical obstacle to Amodei's proposal gaining industry-wide support is the fear that slowing AI research will cede advantage to competitors, according to Sandeep Johri, CEO of cybersecurity firm Checkmarx. This concern is particularly acute regarding competition from companies outside the U.S., given the lack of unified global rules. Johri noted that third-party evaluation can be valuable in projects like OpenAI's "Dawn Project" and Anthropic's "Glasswing Project," which allow partner organizations to access, test, and provide feedback on models. Partners "can strengthen security before the technology is widely released," and Checkmarx is a participant in the Glasswing Project. However, he added that these projects currently focus on cybersecurity and software; addressing broader AI safety risks will require experts from various other fields. The potentially catastrophic consequences if AI spirals out of control amplify the urgency. "When a crisis hits, things escalate very quickly," he said. "I can't say the crisis is here now, but this topic certainly warrants serious discussion."

Amba Kak, co-executive director of the AI Now Institute, stated via email that the AI industry urgently needs enforceable government regulation to prevent harm and ensure safety standards across applications. Kak, a former senior AI advisor at the Federal Trade Commission, proposed that "companies that fail to meet standards should be denied market access, and their executives should face the highest level of legal liability. Only then will we have an effective means to check this industry." She believes third-party evaluators cannot replace independent government oversight. "We are confronting decades of recklessness and opacity in this industry. We need a regulatory framework that doesn't depend on the industry's goodwill to operate," she argued. Timnit Gebru, founder of the Distributed AI Research Institute (DAIR), contends that many existing laws already apply to AI, and suggests that the "existential risk" narrative is overstated. "Existing laws do not make an exception for AI. Fraud is still fraud, and discrimination is still discrimination."