AI safety in further turmoil after Trump’s meeting with tech leaders

After President Donald Trump had lunch with AI leaders on Tuesday, the big artificial intelligence companies are back on track and keeping themselves under tight control.

President Trump entertained executives from various countries. alphabet, meta, space x, Nvidia, Palantir, humanity, OpenAI The topic of AI safety has been a heated national debate for three weeks, with leading industry researchers and executives warning of the extreme threat posed by the most powerful models.

But while many policymakers and industry experts have called for government regulation following concerns expressed this month by Anthropic CEO Dario Amodei and others, President Trump has firmly opposed such a move, calling concerns about AI a “hoax.”

On Tuesday, President Trump brought leaders to the White House to sign a two-page document, which he shared on his Truth Social account. The title is “The White House Agreement on Superintelligence: A Joint Commitment on Frontier Responsibilities.”

The document said the group believes “all companies have a responsibility to develop their technology securely in a way that builds trust with customers and the public.” Signatories include those who have argued that their technology cannot be trusted without concerted security measures.

“Two weeks ago, you would have seen pretty much the same CEO saying it should be regulated,” Tusk Ventures CEO Bradley Tusk said in an interview Tuesday on CNBC’s “Closing Bell Overtime.” “It’s not because they fundamentally believe in the power of the state. They want something that equalizes each other.”

The four action items outlined in the document call on companies to establish “robust” internal oversight of their models, establish an internal team to ensure controls are working, partner with external auditors and evaluators, and appoint an independent committee of the board to oversee the internal team.

“Over time, it may make sense to codify these steps into laws and regulations,” the final paragraph begins. Asked by reporters outside the White House whether the rules were binding, Trump said: “They are morally binding.”

The document left the door open for further regulation, but House Speaker Mike Johnson’s comments make Congress’ push for AI surveillance much less likely. While multiple bills to rein in AI have been proposed in both houses of Congress, and power could shift in the midterm elections, President Trump maintained his position that the industry should be left “exactly where it is.”

President Trump did not just hold the event to promote the new agreement. Separately, he said he had signed a document to “officially” change AI’s name in an effort to push through the rebranding after crowdsourcing a new name on social media last week.

Tech companies are unlikely to switch their use of AI acronyms anytime soon, but their leaders were quick to express support for Trump’s broader efforts after lunch Tuesday. This has been a consistent theme of President Trump’s second term, as American companies, particularly the tech industry, seek to escape the president’s crosshairs.

Meta CEO Mark Zuckerberg told reporters that the deal is intended to “give the American public and our customers confidence that this technology will work as we intended.”

“We don’t think this is all we need to do,” Zuckerberg said. “This is a start and means the whole industry can come to an agreement.”

Compared to colleagues like Amodei and OpenAI CEO Sam Altman, along with Nvidia CEO Jensen Huang, Zuckerberg has been more of a self-regulator. Days after Amodei published an explosive essay earlier this month urging AI labs to “match the pace of the frontier,” Zuckerberg wrote in a post on X that “trust and collaboration are rapidly becoming the most important features that differentiate agents and models.”

Amodei, who has a rocky relationship with the White House but had dinner with President Trump late last week, made a fairly bland comment to reporters after the signing: “If we do this right, if we work with the president and everyone here, we can win safely.”

Alphabet’s Sundar Pichai called the moment “historic and momentous.”

But calls for regulation are likely to grow louder as concerns grow over high-profile security breaches and public sentiment toward AI companies clearly worsens in the run-up to November’s midterm elections.

According to Verasight’s September national survey, 63% of respondents said AI should be slowed down, and only 5% said it should be sped up.

Sen. Mark Warner, D-Virginia, said Tuesday he introduced legislation that would require “rigorous testing and evaluation” of cutting-edge AI technologies and strengthen security and reporting “before the most powerful models are deployed.”

“The companies building the most powerful AI systems are warning us that technology is advancing faster than our safeguards,” Warner said in a release. “The president’s response? Rename artificial intelligence and tell the companies that develop it to self-regulate. But changing what we call artificial intelligence does nothing to address the very real risks associated with increasingly powerful AI systems.”

—CNBC’s Kevin Breuninger, Samantha Subin and Lora Kolodny contributed to this report

WATCH: Top AI and technology executives gather at the White House