The Biden administration today announced a bold and controversial new export control plan designed to prevent advanced chips and artificial intelligence models from falling into the hands of adversaries like China.
The administration’s new “AI Release Act” divides the world into countries that are allowed relatively unrestricted access to America’s most advanced AI silicon and algorithms, and countries that require special permits to access the technology. The rule, which will be enforced by the Commerce Department’s Bureau of Industry and Security, also seeks to restrict the movement of the most powerful artificial intelligence models for the first time.
“The United States is currently a world leader in AI, both AI development and AI chip design, and it’s critical that we continue to do so,” said US Commerce Secretary Gina Raimondo ahead of today’s announcement. Keep the shape.
The list of trusted countries is UK, Canada, Australia, Japan, France, Germany, Belgium, Denmark, Finland, Ireland, Italy, Netherlands, New Zealand, Norway, Republic of Korea, Spain, Sweden and Taiwan.
Under the law, companies from other countries not subject to arms control can obtain up to 1,700 of the latest artificial intelligence chips without special authorization. They can apply for a special license to acquire more chips, build very large-scale data centers using US technology, or access the most powerful “weight” package models made by US companies. Companies will be required to have sufficient physical and cyber security to obtain a license.
Supply chain activities, including the design, manufacture and storage of chips, will be exempt from this rule. The government says the law also does not restrict open-source AI models such as Meta’s Llama.
Arms-embargoed countries such as China, Iran and North Korea are currently barred from acquiring advanced chips. The new law for the first time limits their access to advanced models.
“Semiconductors that have power [AI] And model weights, as we all know, are a dual-use technology. “They are used in many commercial applications, but they can also be used by our adversaries to run nuclear simulations, develop biological weapons, and advance their armies.”
However, the rule is sure to spark controversy, as it may curtail international sales of AI at a critical moment for the industry. It’s only a week before Trump’s inauguration. The ruling sets a 120-day advisory period, meaning the Trump administration is expected to listen to comments, perhaps amend the law and then implement it.
Nvidia, the world’s leading maker of artificial intelligence chips, called the law “unprecedented and wrong” in a blog post. While these laws are disguised as an “anti-China” measure, they do nothing to strengthen US security. Instead of reducing any threat, Biden’s new rules will only weaken America’s global competitiveness and undermine the innovation that has kept the United States at the forefront.
The United States currently restricts exports of advanced artificial intelligence chips to China, a key geopolitical rival, but companies there have been able to build cutting-edge algorithms using computer clusters located in other countries. Under the new law, China cannot build so-called cross-border artificial intelligence models in other countries affected by the law.