Anthropic’s newest AI model, Mythos, is so good, the company says, at uncovering security vulnerabilities that it's too dangerous to release to the public. Anthropic shared a preview version with a select group to help patch the holes that Mythos finds.

But the prospect of a super-hacker AI system is still sending some business leaders into a panic. Marketplace’s Meghan McCarty Carino reports.

Earlier this month, a group called ShinyHunters took responsibility for a hack on the education platform Canvas, which is used for coursework at colleges. In a letter posted online, the group threatened to leak data it took from the platform, including billions of private messages between students and teachers. Canvas was also temporarily unavailable, disrupting students’ ability to do their work.

Then, last week, Instructure, which makes Canvas, said it had reached a deal with the hackers, that the data had been returned and all copies destroyed. Marketplace’s Stephanie Hughes asked Rachel Tobac, CEO at Social Proof Security, what we know about the deal.

As of mid-2025, about a third of newly published websites were generated by artificial intelligence. That’s a massive increase from just three years before when the number hovered around zero.

The AI written text provides fewer diverse viewpoints and is generally presented in a cheerful manner. That's all according to an early study out last month.

Marketplace’s Stephanie Hughes spoke with one of the study’s authors, Stanford University researcher Maty Bohacek, about how AI is changing the nature of the internet.

Hydropower is one of the oldest forms of energy creation, and uses the flow of rushing water to create electricity. The province of Québec, Canada, has historically had an abundance of it and, later this spring, will start supplying hydropower to New York City.

This is the second big hydropower line to link Québec and the Northeast U.S. this year, after a line to Massachusetts came on board this winter.

Right now, the power is going one way, from Canada to the U.S., but some are thinking that the U.S. could eventually sell renewable wind energy back to Canada and let that country hold on to its hydropower for when it's really needed, creating a “regional battery,” says Marketplace's Henry Epp, who’s been reporting on this.

This week, WhatsApp is offering private chats with its AI. Plus, Princeton will now hold supervised exams after a rise in AI-fueled cheating.

But first, Google’s self-driving car division Waymo opted to do a voluntary recall of 3,800 of its robotaxis. This comes after a Waymo drove into a flooded road in San Antonio, Texas. The car was empty and no one was hurt, but Waymo is now updating its software to address how its cars deal with flooded roads.

Will Oremus at The Atlantic joins Marketplace’s Stephanie Hughes for these stories.

Check out our YouTube page to watch more episodes of “Tech Bytes.”

Voice cloning is the use of artificial intelligence to generate a clone of a real person’s voice, imitating the sound, when they pause and what words they typically emphasize. And it can be hard for people to identify voices as being AI-generated.

Research last year from UC Berkeley professor Hany Farid, an expert in digital forensics, found that people correctly identify a voice as AI-generated only 60% of the time.

Marketplace’s Stephanie Hughes spoke with Farid about the rapid sophistication of audio deepfakes, why it's so hard to tell the difference between a real voice and an AI-generated one right now, and some tips to help you spot voice clones.

In order to write her new book “I Am Not a Robot: My Year Using AI To Do ‘Almost’ Everything," journalist Joanna Stern decided to invite artificial intelligence into every aspect of her life — including her family life.

She has a wife and two sons. On their spring break, she took them to Phoenix, where it's easy to hail a driverless car. They rode in a bunch of them, including one that totally freaked out.

She brought home an AI-powered toy (which her four-year-old quickly tired of), and says she realized her kids will "grow up never knowing a world without computers as smart as them.”

Marketplace’s Stephanie Hughes spoke with Stern about how she hopes her children will navigate that world.

Imagine if you invited robots - smart ones or “smart-ish,” at least - into every aspect of your life. Your emails and texts are all composed by an AI, the bots look at a photo of what’s in your fridge and figure out what you can make for dinner. They even become emotional support, providing advice and sometimes companionship. Journalist and founder of media company New Things, Joanna Stern, decided to try this and she wrote about it in her new book “I Am Not a Robot: My Year Using AI To Do Almost Everything.” Marketplace’s Stephanie Hughes spoke with Stern about how AI did and didn’t help her and ultimately what she sacrificed by inviting AI into her life.