The Download

Illustration: Morning Brew Inc., Photo: Getty Images

TL;DR: AI will be banned in the largest public school system in the country starting next week—unless you’re in high school. But New York City’s one-year ban comes with some asterisks, and it’s unclear how it’ll be enforced.

What happened: You may remember that Kentucky middle school that recently went viral for the mistake-ridden AI-generated materials it gave students—including maps with mangled state names like “Lookoong” (that’s Louisiana, in case you don’t speak AI).

It might be a sign of just how routine AI use has become in education—but now the nation’s biggest public school system is fighting against that tide. The one-year moratorium, announced today, will block about 600,000 NYC students from accessing most AI tools, like chatbots and AI tutors, at school.

These restrictions also try to dial back students’ tech exposure more broadly and cover how teachers can wield AI, too. Some of the details:

• Companion chatbots are banned for all grades.
• AI will only be allowed in specific cases even in high school—like career prep or to improve AI literacy.
• Students won’t get individual devices (like laptops) until they reach the third grade.
• And for teachers: AI for lesson planning or translating teaching material is A-OK—but not for grading or counseling students, among other things.

Partial credit: While NYC’s new policy is a notable move to rein in AI at school, some parents and advocacy groups don’t give it a passing grade—they wanted a broader, two-year moratorium on all AI use, which would cover every student as well as schools’ spending on AI tools.

Almost a quarter of kids between 9 and 17 years old use AI every day, according to a recent Common Sense Media survey. And roughly three-quarters of both middle school and high school educators say they use AI in the classroom at least weekly, per an IBM/Morning Consult survey (while about 45% of elementary educators say the same).

Easier said than banned: There are a lot of open questions around how these AI restrictions will work in practice. The benefits of school cellphone bans, for example, have been pretty muddy.

And it’s one thing to block tools like ChatGPT on school networks (NYC did just that in early 2023, then lifted it months later)—but there’s an elephant in the room that classroom bans don’t touch: kids using AI to do their homework.

Bottom line: This new AI policy takes effect next week, when NYC public schools are back in session. A few other US school districts have enacted similar no-AI rules, but the verdict is still out on whether they’re well enforced and actually improve learning. —WK

      

Sponsored By PwC

What’s possible with applied AI? A little while ago, PwC sat down with AT&T at Mobile World Congress to understand their story and success not just with AI but also with staying ahead of the curve with the challenges, trends, and, in turn, opportunities that come their way.

Now that the “what” has been explored, it’s time to tackle the “how” even more deeply.

For AI investments focusing on maximizing ROI, AT&T discusses their AI transformation, leadership working in lockstep with organizational goals, and honing in on AI use cases that matter—not just those that the crowd’s chatting about.

Learn more about AT&T’s AI implementation plans.

• 💊 A new drug for advanced pancreatic cancer just won FDA approval. The decision rested on a late-stage trial of about 500 patients previously treated for metastatic PDAC (the most common form of pancreatic cancer). Patients on the drug survived twice as long as those on chemotherapy and lived longer without signs of cancer progression. Pancreatic cancer has been historically hard to treat—the five-year survival rate is only around 13%. The drug is approved for adults in the US whose cancer has progressed after at least one prior treatment (like chemo).
• 🔭 NASA launched its massive “discovery machine.” The Roman Space Telescope, which launched on Sunday, will “give the Earth a new atlas of the universe,” according to NASA. It’ll help astronomers discover billions of galaxies and tens of billions of stars, as well as thousands of rapid (and thus elusive) supernovas—exploding stars—to track the dark energy that fuels the expansion of the universe. Understanding dark energy and dark matter will help scientists better map what the universe is made of and the movements across the cosmos.
• 🧠 AI helped remove a patient’s brain tumor for the first time. Researchers trained an AI tool on hundreds of videos of tumor removals on the pituitary gland (a tiny organ at the base of the brain) and mapped the vessels and nerves to help it recognize the structures that matter most. Instead of relying on scans prior to the surgery, the surgeons used the tool to analyze a live video feed of the patient’s brain and flag the structures to avoid, marking the safest route to the tumor and areas to cut. The team hopes to develop a kind of ChatGPT for surgeons to have an extra expert on hand. —LC

Sponsored By Currex

Your shoes called. They need backup. Europe’s best-kept insole secret is now America’s most-bought one. CURREX insoles are engineered by German sports scientists to fix how your foot actually moves—not just cushion it. Built for running, soccer, football, hoops, and whatever fall throws at you. Slip them in. Feel the difference before your first sprint.

The Zeitbyte

America’s most anti-AI profession

Morning Brew Design

The AI industry has a very formidable enemy—and it’s not the regulators trying to put a leash on it. Meet insurance claims adjusters, America’s most strident fighters against AI, if Glassdoor complaints are anything to go by.

After they clock out from a day of sifting through mountains of paperwork and dealing with irate claimants, their preferred extracurricular activity is apparently trashing AI on the job review site, per Wired. Glassdoor research found that when adjusters bring up AI on the platform, they’re badmouthing it 98% of the time. (Across all jobs, meanwhile, the proportion of negative mentions of AI is a comparatively sunny 53%.)

The sheer “we hate AI” consensus in this specific role might be surprising, but it’s not rocket science to figure out why. Employment among insurance claims adjusters fell a whopping 21% between May 2025 and May 2026, per Bureau of Labor Statistics data. Adjusters are facing a growing number of job-snatching adversaries—like “AI Jim,” a chatbot that insurance company Lemonade says handles 96% of the initial reports customers file now. And the fatal blow: For a group of famously detail-obsessed people, AI’s tendency to hallucinate is a cardinal sin—and a giant liability that’s probably an insurance professional’s worst nightmare. —WK

Chaos Brewing Meter: ☕☕/5

• Uber is cutting 10% of its workforce—around 3,300 jobs—to reallocate spending to its “ride-sharing, delivery, and robotaxi businesses,” per Bloomberg. To which we say, what else is there?
• This $40,000 solar-powered car looks like a sci-fi dune buggy and is supposed to be the first one “capable of completing the average American’s daily commute on sunshine alone.”
• Google doesn’t have to sell off parts of its business, a judge rules. It does have to make some changes, though.
• Is it an Allbirds? Is it a plane? No, it’s GoPro making an…interesting pivot.
• There’s a new Fable in town, and Anthropic says it costs around 25% less than the OG for “typical workloads.”
• Toyota’s version of Tesla’s Full Self-Driving system is coming to its cars in 2028—with a few differences.
• Read John Ternus’s first memo to Apple staff as CEO.
• You can now use this tool to check whether Claude helped create or edit a file (but it has some limitations).

^{*A message from our sponsor.}

This time last week...

Readers’ most-clicked story was about ChatGPT Work. It can now do what on your computer or browser?!