Fortune colleague Sharon Goldman sat down with the Dell founder and CEO in New York for a conversation about, among other things, “sovereign AIs”—that is, AI developed and controlled by nations or governments.

“They want their culture, their language, and their beliefs, instantiated within their own AI,” he said. “They don’t want this data to go off somewhere else. They don’t want it to be trained by something from California.”

Food for thought amid our AI arms race. Today’s news below. —Andrew Nusca

Want to send thoughts or suggestions to Data Sheet? Drop a line here.

Supreme Court agrees to hear challenge to TikTok ban

The U.S. Supreme Court said Wednesday that it would indeed take up the challenge to the law that could ban TikTok in the U.S.

The high court will hear arguments on January 10, expedited to allow the court to consider them before the law takes effect on Jan. 19, one day before President-elect Donald Trump takes office. That means the Biden administration will present the government’s case.

The core issue is whether the law—which relies on national security grounds to force TikTok’s Chinese parent company, ByteDance, to sell the social media platform to an approved buyer—violates the First Amendment. 

In their SCOTUS request, TikTok’s attorneys described the service as “one of the most significant speech platforms” in the country. TikTok argued in May that a forced sale “is simply not possible.”

But in a Dec. 6 ruling, the appeals court determined that U.S. concerns about Chinese government influence was a “well founded” and “compelling” national security interest. 

European regulators tell AI firms how to avoid big GDPR fines

AI companies have been waiting for Europe’s privacy regulators to agree on how they can stay on the right side of privacy laws, and now they have their wish—maybe.

The bloc’s national watchdogs issued their AI opinion today, making it clear that AI companies have high hurdles to jump if they want to comply with the General Data Protection Regulation (GDPR).

The big problem is the data that the companies use to train their models, which is probably full of personal data (a concept that in Europe means anything you can connect with an identifiable person.)

If that information wasn’t legally collected (it probably wasn’t) and the companies can’t strip it out of their models (they almost certainly can’t) then they could be in real trouble.

GDPR violations can earn fines as high as 4% of global revenues, and the regulators also made clear that they may in some cases require companies to delete their AI models. —David Meyer

Foxconn in talks with Nissan shareholder Renault

The company famous for making iPhones is reportedly in talks with Nissan’s largest shareholder, Renault, about its willingness to sell its shares in the Japanese automaker.

That means Taiwan’s Hon Hai Precision Industry—better known as Foxconn and the world’s largest contract manufacturer of electronics—could complicate reported talks of a merger between Nissan and Japanese peer Honda. 

If Nissan were to engage Foxconn, Honda would end their software partnership, according to a Nikkei report. (Renault’s 36% stake in Nissan derives from a longstanding alliance between the two carmakers and Mitsubishi Motors.)

What does Foxconn have to do with cars? Two words: Electric vehicles. 

In recent years the company has acquired a U.S. manufacturing plant, established an EV supply chain (think chips and batteries), developed an EV software platform called Mobility in Harmony, and unveiled vehicle prototypes in partnership with automaker Yulon Motor as it tries to work more of its technology into more EVs around the globe.

Google contract staff reach union deal banning keystroke monitoring

A group of Google contract workers have reached a union contract with their employer, Accenture, that secures protections on remote work and workplace surveillance and sets a precedent for the many other contract workers in Google parent Alphabet’s orbit.

The collective bargaining agreement between Accenture and the Alphabet Workers Union covers about 25 people who had, among other things, trained generative AI answers for Google’s search products.

The agreement guarantees permanent remote work options, creates a committee where management must consider workers’ input on software, and prohibits Accenture from monitoring their keystrokes or mouse movements. It also provides six weeks of paid time off to look for new positions when there are job cuts.

The workers publicly launched their union drive in June 2023, arguing that Alphabet was a “joint employer” with enough control over them to be legally liable. The National Labor Relations Board agreed, but Alphabet has refused to enter negotiations, arguing that the workers are not its staff. 

The dispute may be resolved in federal appeals court or by the NLRB taking a new stance under President-elect Donald Trump.

More data

—OpenAI announces 1-800-ChatGPT, reminding millions of people that smartphones can make calls.

—The Pentagon says it’s not behind the mysterious drones over the northeastern U.S.

—U.S. weighs ban on TP-Link routers. The Chinese-made hardware is an Amazon bestseller.

—Chime files for IPO. The fintech startup—er, “neobank”—was last valued at $25 billion.

—Epic Games board members resign amid antitrust scrutiny. Two of them also sat on the board of China’s Tencent, which owns a minority stake in Epic.

—MacKenzie Scott has given away $19 billion in the last 5 years. Jeff Bezos…has not!

Endstop triggered