Is ‘self-regulation’ enough for AI? Trump says yes as lawmakers demand rules
by Kerry Picket, Lindsey McPherson · The Washington TimesPresident Trump said he agreed with more than two dozen of the nation’s top AI leaders Tuesday to rely on “tremendous self-regulation” for the industry, turning aside calls for tough rules on the burgeoning technology.
Emerging from a closed-door luncheon at the White House, Mr. Trump said he and House Speaker Mike Johnson had a “very productive, extremely friendly” meeting with tech titans such as Anthropic CEO Dario Amodei, OpenAI President Greg Brockman and Nvidia CEO Jensen Huang.
Mr. Trump said the U.S. has “a very big lead and we’re going to keep our lead” in the industry, adding that some say AI is “bigger than the Industrial Revolution.”
People were interested in these podcasts
“There’s a belief that there should be tremendous self-regulation,” the president told reporters. “We automatically have regulation with … the FBI and all of that.”
The White House and the AI developers signed a “Joint Commitment on Frontier Responsibilities” that Mr. Johnson described as “voluntary.” Mr. Trump pointed to “tremendous self-policing” when asked about penalties if AI companies violate the commitment.
The single-page document identifies “four layers of controls and audits” the AI companies are expected to implement.
SEE ALSO: No more faxes: Trump unveils America.gov, a one-stop shop for government info
The first two describe “robust internal controls” for monitoring AI models during training and deployment to prevent unintended hacks and other unauthorized uses that could pose “cybersecurity, biosecurity and chemical threats.”
The other two layers call for independent audits and for each company to establish a committee under its board of directors to review both internal and external safety reports and “ensure any issues identified are remediated.”
Advertisement Advertisement
The joint commitment was signed by Mr. Trump, Mr. Amodei, Mr. Brockman, Mr. Jensen, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg and Elon Musk, representing the AI division of his social media company X.
The document specifies that the AI companies should meet regularly “to establish standards and best practices to improve the safety of their systems.”
Mr. Trump said he will name a White House AI czar within the next few days. That person will likely have a role in overseeing the execution of the agreement.
Other industry executives who participated in the meeting but did not sign the joint commitment included Palo Alto Networks CEO Nikesh Arora, Amazon founder Jeff Bezos, Altimeter CEO Brad Gerstner, ServiceNow CEO Bill McDermott, Micron CEO Sanjay Mehrotra and Palantir CEO Alex Karp.
SEE ALSO: ‘We will have somebody extraordinary’: Trump to name AI czar within days
Vice President J.D. Vance, Treasury Secretary Scott Bessent, White House counsel Will Scharf and White House Chief of Staff Susie Wiles were among the administration officials who attended.
Advertisement Advertisement
Tuesday’s gathering came amid doomsday warnings about AI and predictions of rogue models hacking into systems, with some industry leaders cautioning that the technology in the wrong hands could become a threat to civilization.
Many lawmakers, particularly Democrats, are pushing for Congress to approve laws setting AI safety standards.
“Voluntary is futile and dangerous,” Sen. Richard Blumenthal, Connecticut Democrat, told The Washington Times.
Big Tech has proven with social media that they cannot be trusted to self-regulate, he said. “There needs to be some governmental oversight and scrutiny, so that there are standards, uniform and protective, and some means of enforcing them,” he said.
Advertisement Advertisement
Mr. Trump says he prefers AI companies self-regulate so the U.S. does not lose a race with China for dominance of AI, which he formally branded “super-intelligence” in an executive order on Tuesday.
“We’re leading by a lot, and we’re gonna keep it that way,” the president said.
Mr. Johnson, speaking on CNBC a few hours before Tuesday’s meeting, also said he would prefer AI leaders to voluntarily agree to oversight and safety guardrails.
“Congress can’t go in and just, you know, take a knee-jerk reaction to overregulate this because the frontier labs are ahead of where members of Congress are,” he said.
Advertisement Advertisement
Senate leaders, none of whom participated in the White House meeting, are still interested in Congress codifying AI safety guardrails.
Senate Majority Leader John Thune, South Dakota Republican, held what he called a “productive” meeting with Mr. Amodei on Monday evening toward that end. He said the Anthropic CEO wants to play a “constructive” role in helping Congress develop a legislative framework for AI regulation that will allow the U.S. to maintain its technological edge over China.
“He understands the importance of ensuring that we strike the right balance: encourage the innovation but make sure we’re doing it in a safe way,” Mr. Thune said.
The majority leader said Mr. Amodei and Anthropic have provided feedback — “largely positive” — to a bipartisan bill he is working on with Sen. Amy Klobuchar, Minnesota Democrat, and Commerce Committee leaders to provide guardrails for advanced AI models.
Advertisement Advertisement
Senate Commerce Chairman Ted Cruz, Texas Republican, has also met with AI executives who are in town this week and said he hopes a bipartisan agreement will materialize after the election.
Earlier Tuesday he blocked an attempt from Sen. Mark R. Warner, Virginia Democrat, to pass a bill by unanimous consent that would establish the Artificial Intelligence Safety Board under the Department of Commerce to bring together government and independent technical experts, to evaluate AI risks and establish safety and security standards.
The measure would require AI developers to provide the board with access to new models at least 45 days before public release for a safety evaluation. Developers who violate the board’s safety standards would be subject to civil penalties of up to $250,000 per violation, per day.
“Are we really going to trust the industry to self-police on something that goes beyond what is happening?” Mr. Warner said. “We’ve already seen not tens but thousands of examples where these models go beyond what humans want them to do.”
Mr. Cruz said he would be happy to work with Mr. Warner on AI regulation but that he has concerns about allowing an executive branch appointed AI safety board to unilaterally decide the safety standards.
He said that leaves too much discretion for “an ambitious bureaucrat” to abuse, especially since there is little existing legal doctrine surrounding AI.
“Federal standards on advance AI must also address the burgeoning patchwork of state laws,” Mr. Cruz said, referring to a GOP effort to preempt similar state laws, which most Democrats oppose.
Contact the author
Lindsey McPherson
lmcpherson@washingtontimes.com
Follow author updates Follow Click to follow. Manage followed authors
Kerry Picket
Follow author updates Follow Click to follow. Manage followed authors