"Jensen Huang, Mark Zuckerberg vs. Dario Amodei: Heated Debate Over AI Risks at the White House"

A debate over the risks of artificial intelligence (AI) reportedly broke out during a White House luncheon meeting attended by heads of major tech companies.

Nvidia CEO Jensen Huang questioned Anthropic CEO Dario Amodei during a meeting with White House staff held in the Roosevelt Room immediately following a White House luncheon on September 29 (local time), asking why he was issuing such extreme public warnings about the risks of AI models, the U.S. daily The Wall Street Journal (WSJ) reported on September 30, citing multiple sources.

In response, Amodei reportedly stood his ground, telling fellow AI company leaders in attendance that they must be honest with the public about model performance and not underestimate the risks.

However, Meta CEO Mark Zuckerberg also addressed Amodei's safety concerns during the luncheon, stating that such issues could be resolved if the industry thoroughly adhered to its own principles, thereby expressing his opposition to the creation of new regulations, the newspaper reported.

Initially, Anthropic, OpenAI, and Google pursued plans to establish a self-regulatory body similar to the Financial Industry Regulatory Authority (FINRA) model, but those plans fell through after Huang, Zuckerberg, and SpaceX CEO Elon Musk expressed their opposition to President Donald Trump.

Recently, Amodei has repeatedly emphasized AI risks and advocated for a slower pace of development, drawing participation from a considerable number of figures in the AI industry. In contrast, Huang has taken the opposing stance, arguing that speed and safety can be achieved simultaneously.

Zuckerberg has similarly opposed the establishment of a regulatory body, arguing that individual companies should set and maintain their appropriate development speeds on their own without external regulation.

On this day as well, Zuckerberg smiled beside President Trump following the luncheon, stating, "We want to give the American people and our customers confidence that the [AI] technology works as intended," aligning with the AI-skeptical Trump administration.

The WSJ pointed out that although Meta suspended President Trump's social media accounts following the U.S. Capitol riot in January 2021, leading to conflict, the company has since settled related lawsuits, recruited figures close to President Trump, and supported the president's key initiatives.

Unlike the debate over AI risks among tech leaders during their meeting with working-level officials, they refrained from exposing disagreements and kept their rhetoric in check during the official events alongside President Trump.

President Trump and House Speaker Mike Johnson announced that the companies attending this luncheon agreed to join the "White House Superintelligence (SI) Agreement"—a pact stipulating that the industry will regulate itself rather than face the creation of new regulations.

Jay Clayton, Director of National Intelligence (DNI) and the administration's AI chief, evaluated the luncheon, saying, "It was one of the most constructive and collaborative industry meetings I have ever seen."

Amodei also told reporters right after the event that if the AI industry and the government cooperate, "we can win safely," and President Trump praised him as "terrific."

President Trump and Amodei previously held a one-on-one dinner meeting at the White House on September 27, hinting at the possibility of improving the strained relationship between the administration and Anthropic.

(Photo: AP, Yonhap News)
※ Please note: This article was translated by AI and may contain errors.