'Not really dug in': John Thune on Trump's stance on AI guardrails

WASHINGTON, DC: Senate Majority Leader John Thune (R-SD) said private talks with President Donald Trump left him believing Trump is “not really dug in” against AI guardrails.

Thune said he believes Trump wants the tech industry to police itself on AI safety rather than face new government rules.

Thune made the comments during an interview with Axios published on Wednesday, September 23, at his Capitol office after discussing AI safety privately with Trump and Anthropic CEO Dario Amodei. 

John Thune says Trump is open to some AI guardrails

Thune said he spoke with Trump a few weeks ago as the debate over AI safety was becoming “pretty hot,” and came away with a more nuanced view of the president’s position.

“I do think that the threats are real,” Thune told the outlet, while saying some of Trump’s resistance is directed at what he described as public “paranoia and noise around the subject.”

Thune said Trump appears to favor having companies develop their own methods for securing AI systems, with experts providing guidance, rather than having Congress or the administration impose broader requirements.

Trump has publicly called recent fears about AI safety a “hoax.” 

But Thune said he does not believe the president is completely opposed to guardrails.

“I think you have to have at least some sort of a plan and some guardrails around it, which is what I think our bill does,” Thune said.

Thune has worked on an AI bill with Sen Amy Klobuchar, D-Minn, in past years, though the legislation has not yet been reintroduced in the current Congress. 

John Thune's bill would put companies on the hook

The bill would place responsibility on technology companies to mitigate AI risks and require companies to report serious risks involving frontier models, the outlet reported.

It would also require companies to develop plans for addressing those concerns, according to Thune. Klobuchar is seeking additional Democratic co-sponsors, including Senators Chris Coons, D-Del, and Maria Cantwell, D-Wash, before the bill is reintroduced.

Thune said Democrats have two major concerns with the legislation. 

One involves pre-deployment testing, which companies can currently choose to undertake under the bill language. Democrats want the testing requirement to be mandatory, the report cited.

Cantwell also wants the Department of Commerce to have AI rulemaking authority. 

Thune said he worries that could give the executive branch too much power.

At the same time, Thune said lawmakers need to distinguish between genuine AI risks and exaggerated warnings.

“But I do also think that there are people who are... doing a lot of fear mongering,” Thune said.

AI safety debate extends beyond Capitol Hill

Thune said Anthropic CEO Dario Amodei called him to discuss AI safety before making his latest public warning about the technology’s risks.

The two also discussed Thune’s bill, as per the report.

The debate is also appearing in US-China discussions. 

Aalok Mehta, director of the Wadhwani AI Center at the Center for Strategic and International Studies, told CNBC that leaders in both countries are concerned about AI models’ cyber capabilities and autonomous systems escaping containment.

Washington continues to restrict China’s access to Nvidia’s most advanced AI chips, while accusing Chinese AI companies of using “distillation” to train models on outputs from more advanced US systems, as per the outlet. 

Beijing rejects that claim.

Rep Ro Khanna, D-Calif, said he hoped Trump and Chinese President Xi Jinping could reach “some common sense safeguards,” including an agreement against self-improving AI. 

“Why wouldn’t you want to sit down with Xi Jinping and say, 'Let’s have some reasonable agreement?'” Khanna said, the outlet cited.