GM’s Cruise is driving back to driverless
On this day in 2001, Apple released the iPod.
On most digital music players of the day, “finding the songs you want to hear on them is as annoying as setting the digital clock on your VCR,” Fortune’s Brent Schlender wrote at the time. But the iPod was just “Plug it in. Whirrrrrr. Done,” according to Steve Jobs.
In other words, it just worked—a value today’s AI makers might want to keep in mind. —Andrew Nusca
Want to send thoughts or suggestions to Data Sheet? Drop a line here.
GM’s Cruise is driving back to driverless
Emily Elconin—Bloomberg/Getty Images
Self-driving cars are pricey, even when they’re not on the streets. General Motors reported in its quarterly earnings call on Tuesday that its self-driving robotaxi company, Cruise, lost $417 million this quarter.
The glass-half-full take is that GM is running Cruise more efficiently. (This time last year, Cruise posted $807 million in operating losses.) But half-a-billion dollars in losses is still a hefty price to pay, especially when you’re not yet commercially operational.
Cruise has been on a hiatus following an incident a year ago in which a woman who was hit by another vehicle was run over by a Cruise robotaxi, then dragged about 20 feet and severely injured as the Cruise car tried to pull over. Regulators alleged Cruise withheld important details of the incident from them, and Cruise voluntarily pulled its vehicles off the streets across the U.S.
It has slowly resumed testing, rolling out cars with safety drivers for tests in Dallas, Houston, and Phoenix earlier this year. On Tuesday, GM revealed in a regulatory filing that it has resumed “limited” driverless testing in Houston—an update, the company told Fortune, that moves the human safety operator from behind the steering wheel to the passenger-side front-seat, still with full access to controls.
There’s no hard timeline for a relaunch of a paid robotaxi service similar to what Cruise rival Waymo provides in many cities. “We expect Cruise to return to driverless commercial AV operations based on safety and other criteria, but we have not yet determined the timing of such return,” a GM statement said. —Jessica Mathews
Netflix gives up on big-budget video games
In an ominous sign for Netflix’s gaming ambitions, the streaming giant has shut down a studio it launched less than two years ago to create big-budget, multiplatform AAA video games. Team Blue, as the Southern California-based game studio was called, never even released a title in its brief life, according to a Game File report.
Netflix isn’t quitting gaming entirely. It has several other studios working on mobile games based on Netflix shows as well as third-party franchises like Grand Theft Auto and Tomb Raider. Just last week, during Netflix’s third quarter earnings call, executives cited games as a “growth initiative” the company was investing in.
But Team Blue was supposed to uplevel Netflix into the gaming big leagues, with blockbuster titles for PCs, and perhaps even consoles, based on original IP. Netflix had hired industry heavyweights like Chacko Sonny, executive producer at Blizzard, Joseph Staten from Microsoft’s Halo franchise, and Rafael Grassetti of Sony’s God of War series to lead the effort.
All three are no longer with Netflix, according to the report. —Jenn Bryce
The creatives have come out against AI
Thousands of famous creative professionals—including actors, musicians, and writers—have signed and published a succinct statement about what they think about tech companies using their work as free training data for AI tools.
In its entirety: “The unlicensed use of creative works for training generative AI is a major, unjust threat to the livelihoods of the people behind those works, and must not be permitted.”
The biggest, most profitable tech companies in the world wasted no time in responding, objecting to the notion that they should pay to use someone else’s work.
The optimistically named Chamber of Progress—a trade group for tech companies like Apple, Meta, Google, Amazon, and others—declared that any kind of licensing regime for copyrighted work put into AI would “seriously compromise creativity, competition, and existing bedrock copyright law.”
Several courts and the U.S. Patent and Trademark Office are actively weighing what should be considered “fair use” when it comes to generative AI and copyright. But really, the argument comes down to money.
Venture capital firm Andreessen Horowitz admitted as much last year in a letter to the USPTO, saying that “imposing the cost of actual or potential copyright liability on the creators of AI models will either kill or significantly hamper their development.” In other words, without reams of free training data, AI doesn’t have a business model. —Kali Hays
Scandal in the land of semiconductors
Over the last half decade, the United States has slowly turned up the heat on one of China’s most valued tech companies.
Five years ago, the U.S. put Huawei on a trade restriction list for national security purposes. A year later it tightened those rules, leading TSMC—the world’s largest semiconductor company by revenue—to completely stop making chips for Huawei. Two years ago, the U.S. outright banned exports to Huawei.
So you can imagine everyone’s surprise when a tech research firm recently took apart Huawei’s Ascend 910B AI chip—a chip composed of other chips that hit the market in 2022—only to find TSMC inside.
In the corporate version of “not it,” the Taiwanese chipmaker immediately notified the U.S. Commerce Department, which reportedly opened an investigation into the matter. (TSMC had made chips for Huawei’s Ascend 910 before trade restrictions took hold.)
The situation raises several questions. Can China build advanced AI chips without access to products made by leading suppliers? Can the U.S. enforce export controls for products that are in high demand? And does any of this deepen concerns that China could, y’know, just seize the company that makes 92% of the advanced chips purchased by the U.S.?
As an Internet-famous cartoon once said: This is fine. —Andrew Nusca
Why you don’t wanna be an accidental Satoshi
Bitcoin’s creator, who goes by the pseudonym “Satoshi Nakamoto,” is theoretically a fabulously wealthy person due to the Bitcoins that they—whoever they are—still hold.
That makes it rather dangerous to be identified as Satoshi, as recently happened to Canadian developer Peter Todd, who once contributed to the Bitcoin project.
After an HBO documentary claimed he was Satoshi, Todd has had to go into hiding, he told Wired.
“Falsely claiming that ordinary people of ordinary wealth are extraordinarily rich exposes them to threats like robbery and kidnapping,” he said. “Satoshi obviously didn’t want to be found, for good reasons, and no one should help people trying to find Satoshi.”
Cullen Hoback, the filmmaker behind the HBO show, disagrees, saying: “The idea that there’s potentially this anonymous figure out there who controls one twentieth of the total supply of digital gold is pretty important.”
Maybe so, but solid proof would be a good idea before naming names, which wasn’t the case with the documentary. —David Meyer
More data
—Meet the Klarna mafia, aka “founder factory.” Whattaya hear, whattaya say?
—Stability AI releases Stable Diffusion 3.5 Large, improving on the diversity of images generated.
—The iPad mini reviews are in. Software: good! Hardware: meh!
—Anthropic’s new Claude 3.5 Sonnet model can mimic keyboard and mouse input to interact with desktop apps. Gulp.
—Meta suspends celeb jet tracking, including for its CEO, as a ToS violation.
