Senior artificial intelligence executives confronted Anthropic Chief Executive Dario Amodei during a high-stakes White House meeting, exposing divisions within Silicon Valley over how the rapidly developing technology should be governed.
The private clash took place during a smaller Roosevelt Room session following a lunch hosted by President Donald Trump and House Speaker Mike Johnson, according to reporting by The Wall Street Journal.
NVIDIA CEO Jensen Huang was among the executives who questioned Amodei over his public warnings about the potential risks posed by increasingly capable AI models.
Amodei defended his position, arguing that the industry needed to communicate risks transparently, particularly concerning potential cyber threats and disruptions to labour markets as AI capabilities advance.
The confrontation came as the administration sought to present technology leaders as united behind new safety protocols and a voluntary approach to AI regulation.
The disagreement highlights wider differences among leading technology companies over how the industry should police itself without measures that could restrict growth or give excessive regulatory influence to a small group of companies.
Earlier proposals for a self-regulatory organisation supported by Anthropic, OpenAI and Google were abandoned following opposition from Huang, Meta CEO Mark Zuckerberg and SpaceX CEO Elon Musk. They argued that such a structure could concentrate too much power among the industry’s leading companies.
Zuckerberg has subsequently emerged as a prominent supporter of the White House’s light-touch approach to AI regulation. He has advocated voluntary industry principles rather than aggressive government intervention or coordinated efforts to slow AI development.
Meta’s engagement with the administration reflects a broader effort by the company to strengthen its relationship with Washington following years of regulatory and political tensions.
Anthropic, meanwhile, is navigating a sensitive political environment as it prepares for an initial public offering that could value the company at about $2 trillion.
The company has sought to repair relations with the administration following earlier regulatory tensions, including temporary security-related model suspensions and disputes over defence-related policies.
The divisions among the technology executives underscore the challenge of balancing rapid commercial expansion with growing scrutiny over AI safety.
The administration is relying on voluntary commitments and pre-release testing of AI models, while investors and policymakers continue to assess whether self-regulation can withstand the competitive pressures driving the industry’s development.
