Trump’s opposition to AI regulations weakens calls for industry slowdown

After the leaders of human, OpenAI After spending the weekend warning of the dangers of advanced AI and calling for a slowdown in model development, President Donald Trump filled his social media feeds with posts undermining those efforts.

“The only control or ‘guardrail’ that AI needs is a strong and smart (high IQ!) president. America definitely has that!” Trump wrote as part of a slew of posts slamming industry executives and critics.

President Trump’s outspoken opposition to new regulations for AI poses new challenges for model leaders Anthropic and OpenAI. Both companies are trying to grow fast enough to justify near-trillion-dollar valuations, while also responding to growing concerns that their technologies may advance so quickly that they may be beyond human control.

Anthropic CEO Dario Amodei published an essay over the weekend urging AI companies to slow down their development after safety researchers Jacob Coxon and Joe Benton resigned from the lab, citing a common fear within the industry that AI systems could kill all humans.

Amodei proposed a plan to audit and regulate AI without “sacrificing commercial advantages” or compromising U.S. national security. With OpenAI CEO Sam Altman space x Elon Musk took the rare step of publicly agreeing with Amodei on each other’s views.

In a fast-growing industry where federal regulation leaves much to be desired, Amodei called for model evaluators to partner with frontier AI companies and to coordinate between democratic and authoritarian governments “to the extent possible” to establish safety standards across democracies.

Amodei said Anthropic is committed to giving evaluators “employee-like access” to monitor and verify the safety of the company’s models. Amodei cited the third-party evaluator as an example: METR, a nonprofit organization founded by former OpenAI and DeepMind researchers whose mission is to “develop scientific methods to assess the catastrophic risks arising from the autonomous capabilities of AI systems and to enable better decision-making regarding the development of AI systems.”

There are only a few other companies in this space, including FAR.AI, AVERI, and Apollo Research.

Tom Wheeler, former chairman of the Federal Communications Commission and current visiting fellow at Brookings University, said companies have no responsibility to effectively regulate themselves.

“Who decides on standards? Who decides on who does the evaluation? Who decides on enforcement?” Wheeler said. It’s important that those at the cutting edge of AI recognize the “huge risks,” but the latest proposals for industry oversight were so vague they amounted to an “empty vessel.”

Virtually every major U.S. industry, from the media and communications industries that Mr. Wheeler helped oversee, to food and beverage, automobiles, pharmaceuticals and financial services, is now subject to tighter federal oversight. AI, which is driving the US economy, is a complete exception and is largely regulated at the state level. For example, California Governor Gavin Newsom recently signed several AI bills into law, including Adam’s Law, which requires AI chatbot makers to protect minors from harmful interactions.

President Trump, who signed an executive order in June requiring AI companies to voluntarily submit their models to the government before making them public, has staunchly opposed any federal regulation and has said concerns about AI amount to “disinformation.”

“When it comes to AI, has anyone in the history of business seen industry leaders pushing for regulations that, if strongly enforced, would drive them into oblivion and bankruptcy?” he wrote in a Truth Social post on Monday.

Meanwhile, President Trump’s former AI czar, David Sachs, continued to criticize Anthropic’s campaign of what he repeatedly called a “regulatory grab” campaign. Sachs, a venture capitalist who invests in AI startups, said in a post over the weekend responding to Amodei’s essay that the two leading model developers should govern themselves instead of asking the government to do so.

“If you think you should slow down because you’re scared of unreleased models, I support your decision to take responsibility,” he wrote.

Representatives for Anthropic and OpenAI did not immediately respond to requests for comment.

The Trump administration has taken the position that anything that impedes AI development in the United States is in the interests of China, its main adversary and the world’s second-largest economy. Jessica Gee, a senior research analyst at Georgetown University’s Center for Security and Emerging Technologies, said the president has shown a willingness to harm companies that oppose him across the private sector.

Anthropic has already been subjected to such treatment, being blacklisted by the Pentagon earlier this year and then receiving a government order in June to revoke access to two of its models to comply with export control directives.

“There are serious concerns about possible punishment by the Trump administration,” Ji said. “The government is trying to encourage them to continue developing and releasing models without stopping.”

In addition to comments from Amodei, Altman, and Musk, microsoft CEO Satya Nadella wrote Sunday in support of “deliberate pacing,” while Google DeepMind chairman Demis Hassabis wrote to X that “Dalio’s essay points to the right path forward.”

Meta CEO Mark Zuckerberg has been aggressively pushing his company’s latest models while also trying to improve relations with President Trump, but has so far remained silent on the matter.

Mehta did not respond to requests for comment.

Sandeep Johri, CEO of cybersecurity firm Checkmarks, said one of the big challenges in getting industry-wide buy-in to Amodei’s proposal is the fear that if one company slows down AI development, another company might take advantage of it. He said that’s especially concerning when considering non-U.S. companies in a world where rules are not defined.

Giori said that in the case of projects like OpenAI’s Daybreak and Anthropic’s Glasswing, third-party evaluations come into play, giving some companies access to the technology to test it and provide feedback. Joli noted that Checkmarx is part of the Glasswing program, which allows partners to “harden their defenses before the functionality is widely available.”

Giori said these efforts only involve cybersecurity and software, and any efforts that may involve AI safety will require a broader range of subject matter experts. The potential catastrophic risk if things go wrong only heightens the sense of urgency.

“Unfortunately, when a crisis occurs, things happen quickly,” he said. “I don’t know if there’s a crisis right now, but it’s definitely a good discussion to have.”

Amba Kaku, co-executive director of the AI ​​Now Institute, said in an email that the AI ​​industry urgently needs “enforceable government regulation” aimed at preventing harm and ensuring safety standards in the areas where the technology is sold.

“Failure to meet that requirement should mean losing market access and facing the most punitive liability, including to individual management,” said Kaku, a former senior AI advisor at the Federal Trade Commission. “Other than that, it gives us an influence that is fundamentally lacking in this industry.”

He said third-party reviewers are not a sufficient substitute for independent government oversight.

“We are facing decades of impunity and opacity,” Kaku said. “We need a structure that allows us to do our work without relying on the permissions of this industry.”

Timnit Gebru, founder of the Distributed AI Research Institute (DAIR), said there are already many laws governing AI today and claims about existential risks for companies are overstated.

“There are no exemptions for AI in existing law,” she says. “Fraud is fraud, discrimination is discrimination.”

WATCH: Genie is reaching the limits of AI, says Crowdstrike CEO George Kurtz.