A dispute over the dangers of artificial intelligence broke out at a White House luncheon gathering of big tech chiefs, according to accounts of the meeting. In public the executives presented a united front in favor of industry self-regulation, but behind the scenes friction surfaced.
The Wall Street Journal, citing multiple sources, reported on the 30th what happened behind closed doors at the luncheon and meeting held at the White House on Sept. 29. The sources said Nvidia Chief Executive Jensen Huang questioned Anthropic CEO Dario Amodei at the meeting held immediately after the luncheon, asking why he issues such extreme public warnings about the risks of AI models.
Amodei did not back down, telling the other AI company chiefs in attendance that they should be honest with the public about what their models can do and should not understate the risks.
Meta CEO Mark Zuckerberg also pushed back during the luncheon after Amodei raised safety concerns, taking a position against new regulation and saying such worries could be addressed if the industry simply adhered strictly to its own principles. The Journal called it the latest example of the taut tension between Anthropic and the rest of the industry, noting that such tension could block the industry's plans for self-regulation covering model evaluations and internal controls.
Anthropic, OpenAI and Google had initially pushed for the creation of a self-regulatory body modeled on the Financial Industry Regulatory Authority, or FINRA, but the plan collapsed after Huang, Zuckerberg and SpaceX CEO Elon Musk conveyed their opposition to President Donald Trump.
Anthropic Treads Carefully Ahead of Listing, While Meta Basks in Warm Ties
Apart from that argument, the AI company chiefs refrained from airing their differences in public and kept their language measured. They joined the White House superintelligence accord — SI being the Trump administration's term for AI — which calls for industry-led self-regulation instead of new rules.
Speaking to reporters right after the event, Amodei said that if the AI industry and the government work together, "we can win safely," and Trump praised the remark as excellent. Trump and Amodei had also held a one-on-one dinner at the White House on the 27th, hinting at the possibility that the strained relationship between the administration and Anthropic could improve.
Anthropic, which is pursuing an initial public offering at a target valuation of $2 trillion, also needs to repair its relationship with the administration. The company was placed on a blacklist amid a dispute with the Defense Department over the scope of military applications for AI.
In contrast to Anthropic's efforts to mend ties, Meta put on display a warmer relationship with the Trump administration than at any time before. Relations between the two had hit bottom after Meta suspended Trump's Facebook account following the Jan. 6, 2021, occupation of the U.S. Capitol. But since the start of Trump's second term, Meta has worked to repair the relationship, settling the lawsuit over the account suspension and investing in projects the president favors.
Zuckerberg is also said to have played a decisive role in drafting the agreement that emerged from the Sept. 29 meeting. Sources said he communicated frequently with Trump as well as administration officials including Commerce Secretary Howard Lutnick. After the luncheon, Zuckerberg stood smiling beside the U.S. president and said the companies want to assure the American people and their customers that the technology works the way they intend it to, falling in step with an administration that is hostile to AI regulation.
