Good morning. The DeepSeek bounce is real, if Chinese tech stocks are to be believed.
The Hang Seng Tech Index, a benchmark tracking the 30 largest tech listings in Hong Kong, is up 25% from Jan. 13, significantly outpacing the Magnificent Seven tech stocks on an equal-weighted basis. Not too shabby.
Whatever DeepSeek’s innovations ultimately contribute to the discourse, it’s clear that investors like what else they see in China—and are putting their money where their mouth is. —Andrew Nusca
Want to send thoughts or suggestions to Data Sheet? Drop a line here.
In France, Vance outlines the good (AI), the bad (China), and the ugly (regulation)
U.S. Vice President JD Vance had three blunt messages for Europe at the AI Action Summit in Paris on Tuesday: Cracking down on disinformation is bad, over-regulating AI is bad, and Chinese tech is bad.
He attacked the EU’s Digital Markets Act (a tech antitrust law) and its Digital Services Act (covering online content) as “onerous international rules,” echoing the complaints of U.S. companies such as Meta that have fallen foul of those laws. As for the new EU AI Act: “Excessive regulation of the AI sector could kill a transformative industry.”
Vance, who just this week has been attacking the independence of the judicial branch in the U.S., warned that “authoritarian regimes” such as China want to use their cheap tech to “infiltrate, dig in and seize your information infrastructure.”
The U.S. did not sign the summit declaration, which referred to “ensuring [that] AI is open, inclusive, transparent, ethical, safe, secure and trustworthy.” Inclusivity is a dirty word for the Trump administration, so its (pardon) inclusion in the declaration ensured U.S. rejection.
The U.K. also declined to sign, with a spokesperson suggesting that doing so would not be in the British national interest.
“We feel very strongly that AI must remain free from ideological bias and that American AI will not be co-opted into a tool for authoritarian censorship,” Vance told delegates. —David Meyer
Thomson Reuters wins AI copyright 'fair use' ruling
A federal judge on Tuesday said that an AI startup was not permitted by U.S. copyright law to copy content owned by Thomson Reuters to build an AI platform.
It’s the first U.S. ruling on the matter of “fair use” in AI-related copyright litigation.
If you’re unfamiliar with the term, “fair use” is a U.S. legal doctrine that allows the unauthorized use of copyright-protected works in a narrow set of circumstances, e.g. nonprofit educational situations, or by using a small portion of the overall work.
It’s a purposefully gray area that allows, for example, a public high school teacher to display Neil Leifer’s famous aerial photograph of the 1966 Muhammad Ali-Cleveland Williams fight during a classroom lecture without penalty but ensures the photographer is paid when the snap is reproduced on a billboard advertising a retailer’s knockout prices.
Figuring out what “fair” is, well, that’s the catch.
In the world of AI, where models must train on content, determining the “fairness” in fair use is the question that separates tech companies like Google, Meta, Microsoft, and OpenAI from creators like artists, engineers who write code, inventors, filmmakers, musicians, and writers.
The tech companies say their generative systems only study protected works to make new content; the copyright owners say those AI-generated results are a little too familiar and threaten their livelihoods.
In the Thomson Reuters suit, which it first filed in 2020, the media and technology company claimed a legal AI startup called Ross Intelligence reproduced materials—its “headnotes” summarizing case points—from Westlaw, the legal research service it has owned since the 1990s. Though the core information was readily available in court decisions, the case hinged on whether the summaries were protected and whether the startup’s offering competed with Westlaw’s.
It did, the judge found. The decision is expected to influence similar cases clogging the U.S. judicial system as AI systems train on every piece of content they can access. —AN
Workday debuts AI agents to ‘peacefully coexist’ with human workers
Workday is joining the list of tech companies pushing into the hot niche of AI agents.
Workday on Tuesday introduced a hub for businesses to manage large numbers of AI agents, which can be unleashed to handle basic digital chores with minimal oversight by humans. It also debuted a handful of agents specialized in specific job roles, including jockeying payroll, financial auditing, and contracts.
In an exclusive interview with Fortune, Workday CEO Carl Eschenbach emphasized that AI agents will be able to take on more than just one specific, step-by-step task like writing software code, fraud detection, or invoice processing.
Eschenbach foresees AI agents as learning and adding new skills over time, ultimately taking on entire roles in a company as “digital employees” that will “peacefully coexist” with human ones. He dismissed fears that agents would replace human workers, generally speaking, but acknowledged they would make certain roles obsolete.
“Humans are going to be re-skilled,” Eschenbach said. “They’re going to want to do different things. They’re going to be doing tasks that they don’t get to today because they’re stuck doing the mundane.”
In moving into AI agents, Workday joins other enterprise software companies like Salesforce, ServiceNow, Microsoft, and Oracle that see AI agents as the next frontier for their offerings. If they can transform their products into dynamic, automated assistants that never sleep, they have the chance to retain customers, gain new ones, and boost revenue.
Workday’s announcement about its AI agents comes a week after it laid off 1,750 employees, or about 8.5% of its workforce. The layoffs, he said, would help Workday, which has a market cap of $72 billion, better focus on customers’ needs and shift more of its focus onto AI. —Sharon Goldman
More data
—Apple reportedly partners with Alibaba for AI features on iPhones sold in China.
—CATL files for Hong Kong IPO. World’s top EV battery maker could raise more than $5 billion.
—DoorDash beats expectations. Q4 revenue up 25% from the previous year to $2.87 billion.
—Google I/O set for May 20-21. Updates on Gemini, Android 16 expected.
—Shopify shuts down Ye store. Swastika tees were a fraud risk, per internal memo.
—Groq inks $1.5 billion deal with Saudi Arabia to supply advanced AI chips.
—Anduril will make U.S. Army goggles. Will overtake a Microsoft arrangement in a reported $22 billion deal, pending Pentagon approval.
—YouTube is the new TV. TV overtook mobile as the primary device for YouTube views in the U.S., CEO Neal Mohan says.
—Meta in talks to acquire FuriosaAI. South Korean AI chip startup is led by Samsung, AMD alum June Paik.
—From VC to CFTC. Andreessen Horowitz’s Brian Quintenz a likely pick to lead U.S. crypto regulator.
—SoftBank posts $2.4 billion loss. A surprise quarterly drop thanks to valuation losses at Coupang, Didi, others.
—Apple follows Google in renaming the Gulf of Mexico to the Gulf of America for Maps users in the U.S.
Endstop triggered
