Nvidia CEO Jensen Huang on Tuesday slammed a proposed antitrust exemption for artificial intelligence companies that could temper a slowdown in the development of increasingly powerful models.
Huang’s comments in an interview with CNBC’s Jim Cramer come amid a heated debate over the safety of AI sparked by the essay. human CEO Dario Amodei made the announcement on Saturday. Amodei suggested major AI labs work together to “pace the frontier” or intentionally slow the development of advanced models to give companies more time to test and protect. Mr. Amodei said the effort could require government intermediaries and exemptions from antitrust regulations.
Huang agreed that AI companies need to take safety seriously, but argued that they should be able to solve these challenges themselves.
“The fact that we need new laws, new antitrust laws, or new regulations for these companies to do the basic engineering properly before they release their products is completely unnecessary,” Huang said on CNBC’s “Mad Money.” “We have a lot of laws. We have a lot of regulations that govern product reliability and functionality.”
In his essay, Amodei outlined a three-step approach to slowing down model development. We start by assigning a third-party safety evaluator to each laboratory. The second and third steps are democratic, global coordination on safety standards and “limiting the rate of unchecked AI progress,” Amodei wrote. “Some adjustments that affect pacing are legally difficult and will require government support.”
Legal issues arise because the major AI labs are competitors and their models are competing products. Although the Sherman Antitrust Act generally prohibits agreements between rivals that unreasonably restrict competition, certain forms of coordination, such as agreements to limit output, may be illegal. This creates potential legal risks if competing AI developers jointly agree on a time frame for developing or releasing new models.
Huang said model developers have the authority to check the safety of their products before releasing them.
“Safety is an engineering issue. Testing is an engineering issue,” said Huang, whose company counts Anthropic and rival lab OpenAI among its key customers. Nvidia is also developing its own Nemotron family of openweight models. “We need to build the product, test it properly, and if it’s not ready to release, hold on to it and keep testing and keep engineering until it’s ready.”
The AI boom has made Nvidia the world’s largest chipmaker and most valuable company. Efforts to slow the development of more capable models could pose a risk to Nvidia’s future growth.
Anthropic did not respond to CNBC’s request for comment on Huang’s comments.
“AI safety is real,” Huang told Kramer. “Designing products to be safe for use around the world is a reality. Companies need to innovate as quickly as possible, but not so fast that they release unsafe products.”
Fan also pushed back against warnings that increasingly powerful AI models pose an existential threat to humanity.
“We will not die in 2030,” he said. “There are a lot of people around the world trying to get AI right. We’re going to put all kinds of guardrails in place and invent all kinds of technologies for safety and security.”
