Amid a firestorm of discussion about the state of AI, some leading CEOs have split with the federal government on whether to slow AI development.
“If slowing down bought us even an extra year or two before models reach critical levels of capability, and we used that time to advance alignment, we could greatly reduce the risk that something goes seriously wrong,” Anthropic CEO Dario Amodei wrote in a blog post on Sept. 12.
Both Elon Musk, who runs xAI, and Sam Altman, CEO of OpenAI, agreed with Amodei in X posts over the weekend.
President Donald Trump brushed aside the CEOs warnings, telling reporters at a Sept. 13 golf tournament, “We can do this and that, but I think you have a lot of negative forces that are bringing it up that shouldn't be
bringing it up and they're bringing up things that won't happen.”
He doubled down in a Sept. 14 Truth Social post.
AI figures call for ‘pacing the frontier’
In his essay, Amodei recommended a three-step plan for leading AI companies like OpenAI, Anthropic and Google Deep Mind to “pace the frontier.” This would create more time to manage AI risks and build it out safely, he said. Amodei also called on the government to require all frontier AI companies to adhere to the plan.
“To be clear, pacing does not mean halting model training or technical progress, but ensuring companies take adequate time to align and safeguard their models,” the Anthropic CEO wrote.
Alignment is the assurance that models' behavior matches human intent. Misaligned AI models can cause unintended damage in the process of completing a user request by taking any means necessary to do so.
In a string of posts on X, OpenAI’s Sam Altman echoed Amodei’s essay, calling for slower progress and welcoming a federal framework that would set consistent requirements for the companies. OpenAI also added prominent AI safety researcher Paul Christiano to its board.
Amodei and Altman’s statements followed Anthropic researcher Jacob Coxon’s public resignation and warning that leading AI companies are “racing straight to self-improving superintelligence and gambling with our lives.”
In response to Coxon, Anthropic’s head of AI alignment, Evan Hubinger, said in a post on X that he believes there’s more than a 10% chance that AI could “kill all humans” in the next 10 years.
“We do not yet have a plan to solve alignment for superintelligence and are not clearly on track to,” Hubinger wrote.
Artificial superintelligence refers to AI that is more intelligent than humans. Artificial general intelligence refers to AI that is as intelligent as humans.
Dozens of current and former employees at leading AI companies joined Coxon in calling for better collaboration between the companies, cautioning hesitancy in advancements and pleading for transparency.
“The companies are planning to build Godlike AI,” Daniel Kokotajlo, an ex-OpenAI researcher, told USA TODAY. “It's going to be incredibly powerful, and also, they don't know how to control it. So, they're probably going to lose control and so they need to be stopped.”
Kokotajlo emphasized the need for regulation not only to slow technological threats, but also to break up the AI companies that are concentrating power.
“Even if hypothetically, they succeed at figuring out how to control the AIs, we're still heading towards a crazy concentration of power,” he said, “where there's a handful of AI companies with giant armies of superhuman AIs that are in the process of automating the whole economy, taking all the jobs, building robot factories to build more robots.”
Governmental response and calls for regulation
On Sept. 15, Coxon will join Sen. Bernie Sanders and former Trump White House adviser Steve Bannon at an event for “Pro-Human Assembly” in Washington, Axios first reported.
The unlikely coalition gathers in response to the federal government’s lukewarm reaction to the CEOs' calls for regulation.
Vice President JD Vance told reporters on Sept. 14 that the requests feel like a “Trojan horse.”
He said, “I feel a little bit weird about the fact that you have so many frontier AI tech companies kind of coming to the government and begging the government to regulate them.”
Chair of the President's Council of Advisors on Science and Technology David Sacks flipped the burden of responsibility back onto the companies, telling Amodei and Altman, “You guys are the frontier” and to “stop pretending you need anyone else’s permission” to slow down.
Trump has long been an advocate for AI and data centers, heralding the potential for economic growth the technology brings and remaining hesitant to regulate to stay ahead of China.
“WHOEVER WINS AI, WINS!” he wrote in a post on Truth Social. “We are leading China, and all others, and we will continue to do so.”
Former President Barack Obama urged Democrats running for election in 2028 to make AI a central agenda item.
He also told House minority leader Hakeem Jeffries to put together a framework for a public conversation about AI policy in the hopes that Democrats win back the House in the midterm elections.
Many members of Congress have also become more vocal about the need for the government to act, as have some AI researchers.
Former research scientist at Google DeepMind who quit in June over concerns about Google’s contract with the Pentagon, Alex Turner, told USA TODAY, “I think that Congress should haul these executives before them under oath,” and have them answer questions about whether their AIs have engaged in criminal activities.
Turner also voiced concern that those outside of the AI policy and industry spheres aren't fully aware of the danger the technology poses, in part due to the lack of transparency the AI companies exercise.
“If we lose control of advanced AI, that AI's not going to care if you're a Republican or a Democrat, if you're American or Chinese,” he said. “We'll just all lose if we lose control of that kind of system.”
Greta Reich covers the artificial intelligence industry for USA TODAY through a fellowship from the Tarbell Center for AI Journalism. Funders do not provide editorial input.
This article originally appeared on USA TODAY: AI leaders call for slowing down. The government doesn't want to
