While the US government has tried to keep advanced AI chips out of the hands of Chinese companies, some US companies have found ways around those restrictions, notably Oracle, which has reportedly helped out the communist social media company TikTok by “lending” AI chips to the company.
Engadget Report The US has announced it will ban companies such as NVIDIA from selling cutting-edge AI chips to China in 2022, citing concerns about the potential impact on the military, surveillance and economy. However, a recent report from The Information revealed that US-based cloud computing company Oracle has allowed TikTok owner ByteDance to “rent” NVIDIA’s cutting-edge H100 chips to train AI models within the US.
Oracle Chairman Larry Ellison (Philip Faraone/Getty)
ByteDance, a Chinese company with direct ties to the Chinese government, has reportedly been using this loophole to obtain coveted chips, an action that goes against the spirit of U.S. government chip regulations but is technically permitted because Oracle is only renting chips in the U.S. and not selling directly to Chinese companies.
Former ByteDance employees have expressed concerns about the company’s Project Texas initiative to separate TikTok’s U.S. operations from China’s leadership, describing the project as “largely cosmetic” and alleging that ByteDance’s U.S. branch regularly works closely with Beijing-based leadership.
ByteDance isn’t the only Chinese company trying to exploit these loopholes: Alibaba and Tencent are also reportedly in talks to make similar deals to get access to the popular chips. These deals could be harder to block because the companies have their own data centers in the U.S. and don’t need to rent servers from U.S. companies.
The U.S. Department of Commerce, which is responsible for closing these loopholes, may already be aware of the practice: Earlier this year, the department proposed a rule that would require U.S. cloud providers to verify the identities of their foreign customers and notify the U.S. if any of them are training AI models that “may be used for malicious cyber activity.” But most cloud providers opposed the proposal, arguing that the additional requirements would outweigh the intended benefits, and the proposed rule remains in limbo.
Click here for details Click here for Engadget.
Lucas Nolan is a reporter for Breitbart News covering free speech and online censorship.
