Companies are quietly replacing workers with AI—but you won’t find that in the press releases. Instead, layoffs are masked behind terms like “restructuring,” “reorganization,” or “business optimization.”

IBM and Klarna are rare exceptions:

But these companies are the outliers. Most firms are hesitant to link AI and layoffs directly not because it’s untrue, but because being too explicit invites scrutiny. Like what happened to Duolingo after declaring itself an “AI-first” company.

Experts say that’s the tell. Job reductions aren’t about financial stress, they’re about AI doing more for less.

But not everything is running smoothly. AI can handle about 90% of many tasks, but the remaining 10% (judgment calls, nuance, edge cases) still needs humans.

And behind closed doors, some executives are growing anxious that their big AI bets aren’t delivering as promised. When the technology falls short, companies rarely walk it back publicly, but instead “quietly outsource or rehire globally to bridge the gap.”

For a sobering look at the trend in real time, Layoffs.fyi is the tech world’s scoreboard, quietly tracking who’s getting “restructured” as it happens.

So where does that leave us?

➡️ The companies that succeed won’t be the ones who replace humans first, they’ll be the ones that figure out how to effectively combine AI and human intelligence. And for workers, the most powerful move is staying informed, adaptable, and open to re-skilling.

As Harvard instructor Christine Inge put it, by the time AI’s impact on jobs is undeniable, it might be too late to do much about it:

Perplexity built a loyal following with fast, reliable AI search that actually cites its sources, making it a go-to for people who want real answers without the noise. Now, they’re raising the bar again with their newest release: an AI-powered web browser called Comet.

Early reviews of Comet are glowing—one user described it as “a completely new way of spending time online” and said it “felt like I was learning how to use the internet for the very first time again.”

Here are its core features:

AI is already changing how we surf the web, and AI browsers are the natural next step. Comet is currently in desktop beta for Perplexity Max users, but it might be landing on your phone soon. Check out the demo here.

With OpenAI rumored to launch its own browser soon, we’ll get to see how they compare—but in the end, strong competition in AI translates to better tools for everyone.

If Morning Brew isn’t on your radar, you’re missing a sharp take on business news. It’s the perfect companion to the Neural Net’s AI deep dives.

Join 4M+ professionals who start their day with Morning Brew—the free newsletter that makes business news quick, clear, and actually enjoyable.

Each morning, it breaks down the biggest stories in business, tech, and finance with a touch of wit to keep things smart and interesting.

Check it out

With major AI rules set to kick in next month, the EU dropped a framework asking companies to be more transparent, skip pirated training data, and respect creator opt-outs. Meta’s response? No thanks, “Europe is heading down the wrong path on AI.” The tech giant says the code is too murky, too demanding, and risks handcuffing innovation by shrinking usable data and piling on legal headaches.

In a recent interview, Perplexity CEO Aravind Srinivas urged people to “spend less time doomscrolling on Instagram; spend more time using the AIs,” not just to boost their careers, but to stay relevant in a rapidly changing workforce. He warned that AI is evolving faster than society can keep up, and for many, the pace is so exhausting that they just tap out. While some experts predict major job losses, others like Nvidia’s Jensen Huang say AI will reshape, rather than erase work. Srinivas’ final message: “People who really are at the frontier of using AIs are going to be way more employable than people who are not.”

The $500 billion Stargate project pitched by OpenAI, SoftBank, and Trump to jumpstart U.S. AI infrastructure is off to a rocky start, with zero data centers built and behind-the-scenes drama slowing things down. The flashy $100 billion “immediate” investment has fizzled into a single pilot site in Ohio. While SoftBank’s still in the mix, OpenAI isn’t waiting around, locking in multibillion-dollar deals with Oracle and CoreWeave instead. The takeaway: even trillion-dollar AI moonshots run into old-school hurdles like land, power, and logistics.

For the first time ever, AI models from OpenAI and Google earned gold medals at the International Mathematical Olympiad (IMO), a global competition for top high school math students. Both companies’ general-purpose reasoning models solved five out of six problems, signaling a major leap in AI’s ability to handle complex mathematical reasoning at a human level.

What makes this especially groundbreaking is how they solved the problems: using natural language instead of formal mathematical code. Traditional AI models relied on symbolic, rigid input formats, but these new models reasoned through problems the way humans do, by thinking and explaining in everyday language.

This means that the models engage more intuitively with complex tasks, making them better suited to collaborate with people and tackle real-world challenges.

As a Brown University professor explained: “The achievement suggests AI is less than a year away from being used by mathematicians to crack unsolved research problems at the frontier of the field.”

AI + advanced math = a party only Silicon Valley could love.