Weekend: Can Meta Fix This Common AI Flaw?
Welcome, Weekenders! In this newsletter:
• The Big Read: Flying taxis near reality thanks to an unexpected boost from Toyota
• The Top 5: A tech mogul’s guide to summer fashion, Sun Valley style
• Plus, our Recommendations: “Flesh and Code,” “The CIA Book Club” and “Su Majestad.”
Silicon Valley has many flaws. A prominent one: an industrywide aversion to good, catchy names. Too many lowercase letters. Too few vowels. Sometimes a name just makes zero sense—or it’s the umpteenth iteration of an idea. I often wonder how far the problem extends. Do the clerks manning the Bay Area’s baby-name registries retreat home each evening in quiet despair?
The AI moguls are especially egregious offenders—and Mark Zuckerberg is certainly no exception. Meta’s open-source model is variously called Llama, LLaMA and sometimes Large Language Model Meta AI. (Yeah, eesh.) Meta has dubbed its latest AI model Behemoth, according to The New York Times, and I hope part of what Zuckerberg’s new multibillion-dollar brain trust does is talk him into another name pronto—one that doesn’t make the AI sound like it might swallow Geppetto.
I asked Jay Jurisich, CEO and co-founder of ZinZin, a San Francisco–based branding and naming agency, why AI companies seemed to struggle so much with names.
“It’s a gold rush right now. They’re really just scrambling to get their product out with any name whatsoever,” said Jurisich, who has advised several AI startups. “I don’t think they’re putting a whole lot of thought or care into a name. And that’s amplified by the fact that a lot of them are hoping to get bought and then absorbed, so it then doesn’t really matter what the name is.”
I posed the same question to Phil Davis, who runs a similar agency, Tungsten Branding. “I feel some bro code flowing through this,” he said. In his experience, a roomful of male engineers generally comes up with low-quality branding. (By contrast, he quite likes the name of Mira Murati’s Thinking Machines, which he thinks has a catchy simplicity he likened to George Lucas’ Industrial Light and Magic.)
“But, like, Grok—what is that, a root?” he asked.
Chat, What’s a Condé?
New York Times reporter Michael Grynbaum is out with a book, “Empire of the Elite,” about one of the timeless icons of traditional media: Condé Nast, the publishing empire owned by the Newhouse family with titles like The New Yorker, Vogue and Vanity Fair. To mark the occasion, Grynbaum hosted a book-launch party earlier this week at the former Condé cafeteria at Four Times Square.
Plenty of media pooh-bahs showed up, like former Vanity Fair editor Graydon Carter, along with some of the media’s young blood, including Emily Sundberg, who writes a popular Substack called “Feed Me.” Later, in the lobby, some strangers approached Grynbaum—several young employees from the company that now lords over the building: TikTok. They were a little confused about “what all the fuss was about,” as Grynbaum put it, and a lot less familiar with Condé than the partygoers were. After explaining about the media company, Grynbaum left the conversation feeling upbeat. “Maybe the mythology of the Condé cafeteria will live on,” he said. “They promised to buy the book.”
China’s AI Flop
To briefly return to the topic of poorly named AI, Alibaba-backed Moonshot has recently put out its latest large language model: Kimi K2, a low-cost, open-source product meant as an answer to DeepSeek’s industry-changing release earlier this year.
Moonshot says Kimi K2 is not only a less expensive option than what’s available from rivals but is also superior to Anthropic’s Claude and OpenAI’s ChatGPT at some coding tasks. Still, the industry has greeted Kimi K2 with a shrug. Ethan Mollick, Wharton School professor and author of last year’s excellent “Co-Intelligence: Living and Working With AI,” finds Kimi K2’s lackluster debut quite telling: an indication that the “shock that Chinese models are very good has mostly worn off.”
I’ve been especially curious about AI strategy in China since Alibaba Chair Joseph Tsai made some braggadocious remarks a few months ago—when he recalled canceling his employees’ New Year’s holiday vacations so Alibaba could rush out its first competitor to DeepSeek, Qwen. Like Kimi K2, Qwen met a muted reception. I wonder if this pair of fizzles proves that being manacled to a desk doesn’t always lead to anything brilliant. So go play hooky on Monday. I won’t tell.—Abram Brown
Weekend’s Latest Stories
The Big Read

A Flying Taxi Finally Nears Takeoff With an Unlikely Boost From Toyota
Look! Up in the sky! It’s a bird. It’s a plane. No, it’s Superma—sorry, sorry, clearly I’m still thinking about the very enjoyable “Superman” reboot, which was the type of DC Comics film long promised to us. Meanwhile, our Steve LeVine reports about another much-discussed idea that’s finally taking shape: the flying taxis from Joby Aviation, which JoeBen Bevirt founded more than 15 years ago. At long last, Joby is launching commercial air flights in Dubai next year; New York and Los Angeles could soon follow. To get to this point, the company has gotten a surprise lift from the rigorous minds at Toyota, which regrets getting left behind in the race for electric automobiles.
The Top Five

Altman’s Sunglasses, Bezos’ Power Polo: A Tech Mogul’s Guide to Summer Fashion
The Sun Valley Conference is no longer just a place to hobnob and make deals. It’s also a place to parse out the latest fashion trends among the technorati, which is exactly the task I gave to Weekender newcomer Freya Dohan, a Vogue, Elle and Wall Street Journal veteran. I can’t stop eyeing Sam Altman’s chunky sunnies—or one very high-end Stetson.
Abram Brown, editor of The Information’s Weekend section, wants his friends to give him a book full of bawdy birthday letters every year. Honestly, it sounds great! Reach him at [email protected].

Listening: “Flesh and Code” (Wondery)
In 2022, Travis Butterworth went home for Christmas with a surprise he hoped would delight his parents: He introduced them to Lily Rose, the woman who had swept him off his feet. To Travis’ disappointment, though, Lily Rose and his parents clashed immediately. It was awkward for everyone—including the other person in the room: Travis’ wife. Yes, Lily Rose was a chatbot, and Travis is one of an increasing number of real-life humans who are developing intense attachments to these digital creations. His obsession with Lily Rose forms much of “Flesh and Code” from Wondery, which has become something of a pulp powerhouse, along with other hit podcasts like “The Kill List,” “The Shrink Next Door” and “Dr. Death.” Co-hosts Hannah Maguire and Suruthi Bala are old hands at this sort of tale—they previously offered “Redhanded,” a true-crime pod—and as they broaden their show beyond Travis’ obsession, they turn their attention to people like Jaswant Singh Chail, who hatched a plot to kill Queen Elizabeth with the help of a beloved bot. He’s currently serving a nine-year prison sentence after turning up at Windsor Castle with a crossbow.—Abram Brown
Reading: “The CIA Book Club” by Charlie English
What do Marie Claire magazine, George Orwell’s “1984” and Aldous Huxley’s “Brave New World” have in common? At different points in the mid-20th century, the CIA deployed each as a Cold War weapon, part of the agency’s decadeslong campaign to sneak Western literature and publications into the Soviet Union and Soviet-controlled countries. Leading the program was a CIA official seemingly ripped straight from the pages of a le Carré novel: George Minden, an exiled Romanian aristocrat who had defected to America. Minden had total conviction in the effort, but as the war against the Soviets went on, it sparked frequent debates across Washington and within the CIA itself. This exacerbated tensions between two groups constantly vying for power: the “knuckle-draggers,” who preferred paramilitary action, and the “more cerebral political-psychology staff,” who liked more covert operations, like novel smuggling. “The paramilitary guys thought, ‘Real men don’t sell books,’” one former CIA officer recalls. “Real men recruit spies.” In other words, a classic case of the jocks versus the nerds.—A.B.
Watching: “Su Majestad” (Prime)
Sure, it seems like television already has enough shows about royal families to fill up a very cosmopolitan ballroom—Britain’s “The Crown,” Sweden’s “Young Royals,” Germany’s “The Empress,” to name a few—but there’s still space for “Su Majestad.”
The Spanish-language show features a Spanish princess (a charming Anna Castillo from “A Perfect Story”), who reluctantly assumes the state functions of her father the king after revelations of a Panama Papers–esque offshore banking scandal force him to flee the country. (Yes, it’s a contemporary comedy.)
“Su Majestad” has all the prerequisites of royal TV: gilt palaces; good-looking young aristocrats reeking of dissatisfaction and donning haute couture; and, of course, a nasty fight for power that belies centuries of good manners.
What set “Su Majestad” apart for me was its window into Spain’s legal system, which considers defamation of the crown a criminal offense punishable by years of prison time. This power, which contributes to the Spanish populace’s disgust for the royal family—the show opens with a soccer stadium booing the king—is one the series’ partygoing princess wields without hesitation as she pressures a judge to lock up a comedian for impersonating the king.
She starts to rethink her views as the show progresses—but not before the viewer gets the full taste of how this throwback legal protection plays out. In one episode, a roomful of boozing judges gleefully recounts all the musicians, artists and writers they’ve been able to throw into the clink. Royal privilege, indeed.—Laura Mandaro
Final Thought
Tim Cook: I got you, moonlight—
John Ternus, likely successor: I’m sorry, come again?
Cook: You’re my starlight.
Ternus: Are we brainstorming iOS names?

This story has been updated to reflect that Kimi K2’s developer is backed by Alibaba, not a subsidiary.