After lunch with leaders in the artificial intelligence field on Tuesday, leading AI companies returned to square one: continuing self-regulation.
Trump hosted executives from companies including Alphabet, Meta, SpaceX, NVIDIA, Palantir, Anthropic, and OpenAI at the White House. Three weeks prior, the topic of AI safety sparked intense debate across the United States, with leading industry researchers and executives warning that the most powerful models could pose extreme threats.
However, concerns raised this month by Anthropic CEO Dario Amodei and others have prompted many policymakers and industry experts to call for government regulation; yet Trump has consistently opposed such measures, calling fears about AI a “scam.”
On Tuesday, Trump invited these leaders to the White House to sign a two-page document, which included a signed version he shared on his Truth Social account. The document is titled "The White House Superintelligence Agreement: Joint Commitment to Frontline Responsibility."
The document states that the group believes, "Every company has a responsibility to develop its own technology in a secure manner that builds trust with customers and the public." Among the signatories are individuals who have previously emphasized that their technology cannot be trusted without coordinated security efforts.
Bradley Tusk, CEO of Tusk Ventures, said on CNBC’s Closing Bell Overtime on Tuesday: “You saw two weeks ago that nearly all of those same CEOs said they should be regulated. It’s not because they fundamentally believe in the power of the state—they want something that creates balance among themselves.”
The four actions outlined in the document require each company to establish a “robust” internal model monitoring mechanism, form an internal team to ensure the effectiveness of controls, collaborate with external auditors or evaluators, and designate an independent committee within the board to oversee the internal team.
The final paragraph states: “In the long term, it may be meaningful to codify these steps into law or regulation.” When asked outside the White House whether these rules are binding, Trump said they are “morally binding.”
Trump didn't host this event just to promote the new agreement. He also stated that he signed a document formally renaming AI as "super intelligence," aiming to advance this rebranding effort following last week's social media crowdsourcing of a new name.
This is just the beginning.
Although tech companies are unlikely to abandon the abbreviation AI anytime soon, executives quickly expressed support for Trump’s broader efforts after Tuesday’s lunch. This has been a consistent theme during Trump’s second term: corporate America, particularly the tech industry, has sought to avoid the president’s wrath.
Meta CEO Mark Zuckerberg told reporters that the agreement aims to "give the American people and our customers confidence that this technology will work as intended."
Zuckerberg said: "This doesn't mean it will be the only thing we do going forward. It means this is a starting point and an agreement that the entire industry can reach."
Compared to peers such as Amodei and OpenAI CEO Sam Altman, Zuckerberg has consistently favored self-regulation, as has NVIDIA CEO Jensen Huang. Just days after Amodei published a highly influential article calling for AI labs to “slow down frontier progress,” Zuckerberg posted on X, “Trust and alignment are rapidly becoming the most important capabilities distinguishing agents from models.”
Amodei’s relationship with the White House has been rocky, but he had dinner with Trump earlier last week. After the signing ceremony, he told reporters, “If we do this right, if we work with the president and everyone here, we can win safely.”
Alphabet CEO Sundar Pichai described this moment as "historic and far-reaching."
But as high-profile security vulnerabilities escalate and public sentiment toward AI companies declines ahead of the November midterm elections, calls for regulation will only grow louder.
According to Verasight’s September national survey, 63% of respondents said AI should slow down, while only 5% said it should accelerate.
Virginia Democratic Senator Mark Warner said on Tuesday that he has introduced legislation requiring "rigorous testing and evaluation" of leading AI technologies and strengthening safety and reporting requirements "before the most powerful models are deployed."
Warner said in a statement: “The companies building the most powerful AI systems are warning us that the pace of advancement has outstripped our safeguards. What’s the president’s response? Rename it and tell the companies developing it to self-regulate. But changing what we call artificial intelligence doesn’t address the very real risks posed by increasingly powerful AI systems.”
Kevin Breuninger, Samantha Subin, and Lora Kolodny of CNBC contributed to this report.