Breaking

Meta has settled another landmark case around social media addiction and agreed to pay up to $17 billion in fines. It will also implement numerous restrictions for Facebook and Instagram users under 18, including a two-hour daily limit, blocking use between midnight and 6am, and the ability to choose a non-algorithmic feed (hey, we want that too!). Here’s the full list of changes—which it’s calling on TikTok and YouTube to implement, too.

The Download

Morning Brew Inc., Photo: Annette Riedl/Getty Images

TL;DR: Bill Gates, in his own words, might now be the “shrillest voice” on AI risk. He’s done a flurry of interviews (which all dropped today) sounding the alarm on a technology he thinks could destabilize the world, posted a 5,900-word blog post on how it could be the “worst source of injustice”—and says he’s been talking about it with leaders in government and the tech industry, too. The question is, will people listen?

What happened: In a wide range of interviews published today, the Microsoft co-founder sent a dire warning about how AI could drive job loss and bioterrorism. And he’s not alone. Behind closed doors, Gates says tech executives are actually “very worried” about AI—but are downplaying it publicly.

The tech billionaire hasn’t exactly embraced the spotlight lately. He acknowledged to Axios that his past relationship with Jeffrey Epstein (which he called a “huge mistake”) could impact how much weight his voice has on the AI debate. But he’s speaking out now because AI has already crossed critical thresholds that make it dangerous. “I am in a state of shock that I’m sort of the first one saying, ‘This is crazy. This is insane,’” he told Semafor (a comment that Dario Amodei might take issue with).

Pathogens and paychecks: During his media tour, Gates zeroed in on the tech’s potential impact on jobs, saying that there won’t be a job left unaffected because humans will eventually be “vastly inferior to the AI doing the job.” And the other thing keeping Gates up at night: he now believes the risk of bioterrorism is about 50 times higher than the risk of a natural pandemic.

How to avoid the bad outcome: The former tech CEO didn’t only throw out bleak forecasts, though. Gates proposes an AI token tax that would make automating roles less financially attractive—and suggests that certain jobs (like athletes and childcare workers) should be “human reserved.” The billionaire also thinks voluntary AI oversight measures are a bad idea and calls for international agreements around AI monitoring: “Self-regulation on the most dangerous tool ever invented?” he said to the NYT. “No, thanks!”

Bottom line: Gates argues that his message might resonate exactly because he’s an unlikely messenger—an “innovation lover” who’s now deeply concerned. Though the fact that he’s an influential billionaire better known these days for his philanthropy than for Microsoft probably doesn’t hurt. —WK

      

Sponsored By Delight.AI

The hidden cost of legacy software isn’t just migration: It’s paying per seat for a system that limits your team. That’s where Delight Desk comes in.

Delight Desk provides one-click migration from your previous solution while preserving your tickets, history, knowledge base, macros, users, custom fields, and help-center URLs. And the seat fee is $0 forever—teams pay when AI resolves a ticket, not when human agents do.

Here’s how it works. Delight Desk’s AI arrives pre-trained based on historical tickets, with Actionbooks drafted from real resolutions. Then, they’re held for human approval. With a familiar interface, your team needs less training and can move on your timeline.

Get started.

• 💊 This pill promises to mimic the effects of exercise. It imitates a molecule the body releases during hard exercise—which turns down hunger signals in the brain—to curb appetite, according to its creator. Results from the drug’s first clinical trial, conducted on 88 healthy adults, showed that the pill worked without the drawbacks commonly experienced on GLP-1s like Ozempic (such as muscle loss and nausea). The company is also planning to test it against migraines and inflammatory bowel disease—other conditions exercise is known to help with.
• 🍪 Scientists made a cookie out of plastic (yes, it’s apparently edible). The cookie was originally made as part of NASA’s Deep Space Food Challenge (designed to create food systems to sustain long space missions). It’s created by blending PET (the clear polyester plastic bottles are made of) with biomass like corn stalks and leaves. Hot, pressurized water and oxygen break the mixture into microbe-sized bites, and engineered yeasts turn the plastic into proteins, vitamins, and flavorings. Then some baking ingredients and a 3D printer turn all that into a cookie. While the current version is still really expensive to make ($60 for 2.2 pounds), it’s a step forward in addressing natural food supply issues on deep-space missions or in disaster zones. —LC

Sponsored By Catch AI

Ready when you are. An AI agent built exclusively for administrative work? Catch AI built it. This AI agent can monitor your inbox, handle your schedule, coordinate meetings—even manage business travel. Talk to it however you prefer: text, email, Slack, WhatsApp, or phone. It’s available 24/7, proactively getting work done in the background. Check it out.

The Zeitbyte

Tech swag goes Swiss

Vanguart via The Wall Street Journal

It turns out that when you have “we jumpstarted the AI race” kind of money, you can take typical rich-person hobbies to a whole new level. Behold: Sam Altman’s OpenAI-branded wristwatch from Swiss luxury brand Vanguart, which has the phrase “COMPUTE IS DESTINY” engraved on the rotor, per The Wall Street Journal. This custom watch is based on a model that starts out at a cool $180,000—and Altman commissioned seven.

Tech bros and expensive wrist candy aren’t exactly a novel combo (Mark Zuckerberg recently flashed a wristwatch that costs $1.3 million), but it’s less common to see one stuffed full of brand-related easter eggs. Altman’s bespoke creation includes plenty of other details that read as very earnest, very cringe, or both—like the Python code “if AGI.aligned: deploy()” inscribed on the back. And it’s not even the only OpenAI-stamped watch out there: OpenAI’s head of applied research recently showed off a watch that features the company logo on the dial and the phrase “Good Research Takes Time” on the back. ChatGPT might struggle with telling time, but its makers apparently can’t get enough of it. —WK

Chaos Brewing Meter: ☕☕/5

• All those job cuts at Meta were reportedly just the first step—it planned to go way, way further until staff revolted.
• Anddd another OpenAI staffer bites the dust.
• Outsourcing tasks to humans is so passé. At least according to Amazon, which is sunsetting this service after 21 years.
• Errr…ChatGPT Work can now do what on your computer or browser?!
• Rockstar says the GTA VI leaks have been “heartbreaking” and could impact players’ experience of the game after finally breaking its weeklong silence.
• It seems like OpenAI’s spicy chip is all anyone can talk about this week.
• Anthropic says “hold my beer” to SpaceX—it’s reportedly going to tell investors its potential revenue could reach over $30 trillion.

^{*A message from our sponsor.}

This time last week...

Readers’ most-clicked story was about Gwyneth Paltrow and Sam Altman. Um, what exactly are they doing together?