Meta CEO Mark Zuckerberg was the chief architect of the “morally binding” AI accords that emerged from Tuesday’s White House lunch with President Donald Trump, Speaker Mike Johnson, and more than a dozen tech executives, according to five people familiar with the private meeting.
Zuckerberg, with the help of Nvidia CEO Jensen Huang, pitched the so-called AI accords to senior White House officials and some of the biggest names in tech during the luncheon, three of the people said. Both Huang as well as Trump administration officials helped shape the pact ahead of the meeting, according to two of the people.
All five people were granted anonymity because they were not authorized to share details of the discussions.
As part of the voluntary agreement, signatories would commit to the importance of enacting safeguards to mitigate potential risks from advanced AI models, but without a clear enforcement mechanism, it is unclear what, if any, consequences they may face for failing to do so.
The episode indicates Zuckerberg and Huang’s growing influence shaping White House AI policy, and comes as the administration continues to pursue a light-touch approach to regulating the technology. Congress has yet to pass federal legislation reining in AI.
Meta declined to comment. Nvidia and the White House did not immediately respond to a request for comment.
The idea originated at a state dinner for Chinese President Xi Jinping, during which Zuckerberg pitched the idea to Johnson, one of the people said. Shortly thereafter, Zuckerberg spoke with Huang who subsequently helped to garner support for it among leadership at the major AI labs, according to two of the people. Since the state dinner until Tuesday’s luncheon, executives have been in consistent communication to finalize the document.
Semafor previously reported some of the details of the document’s creation.
The accords compel companies to “implement robust internal controls” to monitor AI models’ capabilities, partner with external auditors, and create independent committees to review the auditors’ reports. It has been criticized by some Democrats and safety advocates who say it is unenforceable and relies primarily on mutual agreement.
“The idea isn’t that this is the only thing that we will ever do,” Zuckerberg told reporters outside the White House Tuesday. “It is that this is a start and an accord that the whole industry could come to.”
Trump has been reluctant to impose additional guardrails on the technology, saying that existing laws will be enough to mitigate the risks the technology may pose. “The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades,” Trump said in a post on Truth Social earlier this month.
As the White House has weighed how best to regulate AI, Zuckerberg has favored a lighter-touch approach. In an August phone call with Trump, Zuckerberg opposed a proposal that would create an independent regulatory body modeled on the Financial Industry Regulatory Authority.
Huang has positioned himself as a coalition builder within the industry. On Monday, Nvidia released a new AI security system the company said would help prevent recent autonomous hacks conducted by rogue AI agents that have made headlines.
Other top tech executives, including Anthropic’s Dario Amodei and OpenAI’s Sam Altman, have called for the government to have a heavier hand in regulating the technology and called for a slowdown of its development.
Zuckerberg and Huang, though, have largely dismissed critics who say that AI could pose an existential risk to humanity.
This story originally appeared on POLITICO and is courtesy of the Axel Springer Global Reporters Network, which harnesses the resources of the company’s newsrooms to publish ambitious scoops, investigations, interviews, opinion pieces, and analysis. It allows journalists — including those from POLITICO, Business Insider, WELT, BILD, Onet, and Fakt — to collaborate on major stories for an international audience of hundreds of millions across platforms.
