Google's has introduced new AI Overviews to make searching more intuitive. NPR's Bobby Allyn speaks to University of Pennsylvania professor Ethan Mollick about why some of the results have been inaccurate — and ocassionally absurd.

A new lab analysis conducted for NPR by Arizona State University data scientists shows that OpenAI's "Sky" voice is more similar to Johansson's than hundreds of other actors analyzed.

(Image credit: Angela Weiss)

There is a constant arms race between law enforcement and criminals, especially when it comes to technology. For years, law enforcement has been frustrated with encrypted messaging apps, like Signal and Telegram. And law enforcement has been even more frustrated by encrypted phones, specifically designed to thwart authorities from snooping.

But in 2018, in a story that seems like it's straight out of a spy novel, the FBI was approached with an offer: Would they like to get into the encrypted cell phone business? What if they could convince criminals to use their phones to plan and document their crimes — all while the FBI was secretly watching? It could be an unprecedented peek into the criminal underground.

To pull off this massive sting operation, the FBI needed to design a cell phone that criminals wanted to use and adopt. Their mission: to make a tech platform for the criminal underworld. And in many ways, the FBI's journey was filled with all the hallmarks of many Silicon Valley start-ups.

On this show, we talk with journalist Joseph Cox, who wrote a new book about the FBI's cell phone business, called Dark Wire. And we hear from the federal prosecutor who became an unlikely tech company founder.

Help support Planet Money and hear our bonus episodes by subscribing to Planet Money+ in Apple Podcasts or at plus.npr.org/planetmoney.

Four out of five companies in Europe's largest economy continue to use fax machines. But Germany's parliament has until the end of June to stop relying on the antiquated communication technology.

(Image credit: Holger Hollemann)

Most of the tools tested by researchers at the nonprofit Center for Countering Digital Hate could be used to successfully clone a wide range of voices belonging to European and American politicians.

(Image credit: Elijah Nouvelage)

AI tools have helped the people behind influence operations produce more content, but OpenAI says the operations it disrupted didn’t gain traction with real people or reach large audiences.

(Image credit: Stefani Reynolds)

ChatGPT-maker OpenAI is confronting fresh questions about how seriously it treats AI safety. Former employees and others say the company should not be trusted with governing itself.

Our special series, Body Electric, is back! This time, host Manoush Zomorodi does a deep dive into our headphone habits... because many of us wear them for hours at a time, and all that listening is taking a toll on our hearing.

Manoush speaks with exposure scientist Rick Neitzel, who has partnered with Apple on a first-of-its-kind study to find how our daily listening patterns are affecting our hearing. Neitzel offers advice on safe listening habits that can help protect our ears in the long term.

Later, Manoush takes us into the future of "consumer hearables" and how tech companies want us to never — ever— take our earbuds out.

Interested in joining the Apple Hearing Study? Sign up here.

Binge the whole Body Electric series here. Sign up for the Body Electric Challenge and our newsletter here.

Talk to us on Instagram @ManoushZ, or record a voice memo and email it to us at BodyElectric@npr.org.

Books take a lot of resources to make. Digital readers do, too. What's the more sustainable option? The answer isn't straightforward.

(Image credit: JGI/Daniel Grill)

NPR's Scott Simon speaks with Amanda Fennell, a cybersecurity expert, about CAPTCHA tests that verify if a website visitor is human or bot. They are getting harder.

After the Hamas attack of Oct. 7 triggered Israel's invasion of the Gaza Strip, hundreds of thousands of Palestinians began fleeing from the North of Gaza to the South. As they fled, many Palestinians reported passing through checkpoints with cameras. Israel had previously used facial recognition software in the West Bank, and some Palestinians reached out to The New York Times reporter Sheera Frenkel to investigate whether the same was happening in Gaza.

Science correspondent Geoff Brumfiel talks to Frenkel about how Israel launched this facial recognition system in Gaza late last year with the help of private companies and Google photos.

Read Frenkel's full article.

Want to hear us cover more stories about AI? Email us at shortwave@npr.org.

(Image credit: NurPhoto)

Lawmakers on Capitol Hill recently got a crash course in quantum science. The technology poses a range of threats and opportunities.

The fines and charges come after New Hampshire voters got robocalls from an AI-generated version of President Biden's voice urging them not to vote in the upcoming presidential primary.

(Image credit: Tasos Katopodis)

With IT systems down, staff at Ascension have to use manual processes they left behind some 20 years ago. It's the latest in a string of attacks on health care systems that house private patient data.

NPR's Ari Shapiro speaks with Commerce Secretary Gina Raimondo about a new "global AI safety network."

Many of us wear earbuds for hours at a time, sometimes all day long, and all that listening is taking a toll on our hearing. This episode, host Manoush Zomorodi investigates our headphone habits. She speaks with exposure scientist Rick Neitzel, who has partnered with Apple to create a first-of-its-kind study into how our daily sound exposure and listening patterns are affecting our hearing. Neitzel offers advice on safe listening habits that can help protect our ears in the long term.

Later, Manoush takes us into the future of "consumer hearables" and how tech companies want us to never — ever— take our earbuds out.

Interested in joining the Apple Hearing Study? Sign up here.

Binge the whole Body Electric series here.

Sign up for the Body Electric Challenge and our newsletter here.

Talk to us on Instagram @ManoushZ, or record a voice memo and email it to us at BodyElectric@npr.org.

Bad weather may have been a cause of Sunday's crash that killed Ebrahim Raisi. But mechanical issues, possibly exacerbated by a lack of spare parts due to U.S. sanctions, could also be a factor.

(Image credit: Atta Kenare)

Scarlett Johansson says she was approached multiple times by OpenAI to be the voice of ChatGPT, and that she declined. Then the company released a voice assistant that sounded uncannily like her.

The Seoul summit is a follow-up to last November's summit in the U.K., where participating countries agreed to work together to contain risks posed by galloping advances in artificial intelligence.

Johansson says she was approached multiple times by OpenAI to be the voice of ChatGPT, and that she declined. Then the company released a voice assistant that sounded uncannily like her.

(Image credit: Leon Bennett)

Oversharing can make children vulnerable to identity theft, harassment and predators. To protect their privacy, share a 'holiday card-or-less' amount of data online, says expert Leah Plunkett.

Scientists are testing the limits of artificial intelligence when it comes to language learning. One recent challenge? Learning ... whale! Researchers are using machine learning to analyze and decode whale sounds — and it's just as complicated as it seems.

Curious about other mysteries of nature? Email us at shortwave@npr.org.

A helium leak pushed back a planned launch to May 25. Boeing's program that would shuttle astronauts to and from the International Space Station has been plagued with problems.

Last month, the world narrowly avoided a cyberattack of stunning ambition. The targets were some of the most important computers on the planet. Computers that power the internet. Computers used by banks and airlines and even the military.

What these computers had in common was that they all relied on open source software.

A strange fact about modern life is that most of the computers responsible for it are running open source software. That is, software mostly written by unpaid, sometimes even anonymous volunteers. Some crucial open source programs are managed by just a single overworked programmer. And as the world learned last month, these programs can become attractive targets for hackers.

In this case, the hackers had infiltrated a popular open source program called XZ. Slowly, over the course of two years, they transformed XZ into a secret backdoor. And if they hadn't been caught, they could have taken control of large swaths of the internet.

On today's show, we get the story behind the XZ hack and what made it possible. How the hackers took advantage of the strange way we make modern software. And what that tells us about the economics of one of the most important industries in the world.

Help support Planet Money and hear our bonus episodes by subscribing to Planet Money+ in Apple Podcasts or at plus.npr.org/planetmoney.

NPR's Ailsa Chang talks with Maggie Harrison Dupre, staff writer at Futurism, about her reporting into AI-generated articles appearing on major news publications.

The latest version of ChatGPT has the internet wondering: Was it meant to make it sound like Scarlett Johansson in the movie Her? Its creators insist the model was not based on the movie.

"Russia remains the most active foreign threat to our elections," said Director of National Intelligence Avril Haines, noting that new AI technologies make influence operations easier to pull off.

(Image credit: Win McNamee)

This comes after a yearlong listening tour by a bipartisan working group in the Senate.

(Image credit: Allison Bailey)

The Justice Department is expected to argue that its clamp down on TikTok is about national security, but Constitutional lawyers say there is no way around grappling with the free speech implications.

(Image credit: Michael Dwyer)

AI can conjure the voice or likeness of a dead celebrity with just a few clicks. This opens a host of legal questions about the rights of the deceased and their heirs to control their digital replicas

(Image credit: Raymond Boyd)

Stanford students on a hackathon team have created an AI tool designed to help veterans apply for disability benefits. Can their tool beat the Department of Veteran Affairs' notorious red tape?

A Jesus made of vegetables, bizarre log cabins, products that don't exist. AI-generated images are creating new forms of clickbait and causing some users to doubt what's real.

A new type of traveler is part of the post-pandemic reset at U.S. hotels, along with fewer daily cleanings and pancake-slinging machines.

(Image credit: Amar Patel)

Stanford students on a hackathon team have created an AI tool designed to help veterans apply for disability benefits. Can their tool beat the Department of Veteran Affairs' notorious red tape?

As artificial intelligence seeps into some realms of society, it rushes into others. One area it's making a big difference is protein science — as in the "building blocks of life," proteins! Producer Berly McCoy talks to host Emily Kwong about the newest advance in protein science: AlphaFold3, an AI program from Google DeepMind. Plus, they talk about the wider field of AI protein science and why researchers hope it will solve a range of problems, from disease to the climate.

Have other aspects of AI you want us to cover? Email us at shortwave@npr.org.

Companies in China are using deepfake technology to create avatars of dead relatives and loved ones. Does the technology help or hurt the grieving process?

Here's a summary of NPR's reporting about a purported flyer that was found in a portable toilet at a migrant encampment in Matamoros, Mexico that urged migrants to vote illegally for President Biden.

The National Oceanic and Atmospheric Administration says there have been measurable effects and impacts from the biggest geomagnetic storm in decades.

The newest iPad ad depicts instruments, books and art supplies flattened into Apple's thinnest product ever. But anyone who owns and loves art in any form knows: The practicality isn't the point.

Viral images of the flyer were filmed in portable toilets of a migrant camp in Mexico, and they energized members of Congress. But NPR's reporting suggests the flyer is not what it purports to be.

The San Francisco-based AI juggernaut says it is re-evaluating its policies around "NSFW" content.

The White House wants a twenty-fold increase in geothermal energy production to fight climate change and it's counting on the oil and gas industry for help.

(Image credit: Kirk Siegler)

Rep. Jennifer Wexton of Virginia was diagnosed last year with rare disease that makes it hard to speak. She still advocated for a bill renaming a post office in her district – and the House passed it.

Apple unveils new versions of its iPad at a time when revenue from its devices are falling and it faces growing competition from places like China. Will this refresh help the company?

The high-stakes legal battle could determine the future of the popular app in the U.S. TikTok's legal filing calls the ban law an unprecedented violation of First Amendment rights.

(Image credit: Kiichiro Sato)

Katy Perry and Rihanna weren't at the Met Gala on Monday night, despite the viral AI-generated photos showing them on the red carpet. Here are some tips for recognizing and investigating deep fakes.

(Image credit: Jordan Strauss)

Bumble, known for allowing women to message men first, unveiled new features that allow men to make the first move. Will the change breathe new life into online dating, and the company's stock?

(Image credit: Nikos Pekiaridis)

Modern human life relies on a stable internet connection. But threats to internet connectivity are varied — from underseas rock slides and technical errors to war and geopolitical conflict.