Young children who directly interact with chatbots face new developmental risks, pediatricians and psychologists warn.

Why it matters: Assessing the impacts early could help protect a new generation from becoming guinea pigs for tech we don't fully understand.

The big picture: Most kids under 5 aren't using voice-based AI yet. Young kids already see AI-generated content — but on its own, that isn't necessarily a new problem.

• Low-quality images and videos commonly known as AI slop don't differ that much from the low end of mass-produced kids' media that's been around for decades.

The real concern is interaction with live AI, which can be both powerfully engaging and confusing.

• Imagine "Blue's Clues," but the character answers in real time and adapts to a child's input.

Chatbot use — especially among young children — is still new. There are no long-term studies on its impact, but we do know a lot about how screen time and tech use affect child development.

• The amount of time kids spend on devices is disrupting important developmental milestones, Scott Kollins, psychologist and chief medical officer at family software company Aura, told Axios.
• "We've seen that with older kids. And with kids between 0 and 5, their brains are developing faster, and they're hitting these important social milestones more frequently."
• Interacting with AI chatbots, Kollins warns, has "the potential to be even that much more disruptive" — but right now, "it is all conjecture in terms of this specific type of interaction."

What they're saying: Interaction with generative AI could "fundamentally change the human brain," says Dana Suskind, a pediatric physician and expert on early childhood and early language development.

• Suskind says teenagers and adults are already forming relationships with AI companions. The same could happen with younger kids.
• "The content and experience that kids are exposed to in early years isn't just sort of changing things the same way social media impacted adolescent brains," Suskind told Axios. "It is actually changing the foundational wiring of the human brain."
• "Children naturally anthropomorphize," Suskind wrote in an email, "but with responsive AI, we're entering uncharted territory for how this might shape their developing sense of reality and relationships."

Between the lines: Some child development researchers worry that chatbots could reshape how children learn trust, empathy and connection.

• A small study from 2024 showed that kids ages 3-6 were more likely to trust a robot than a human, even when that robot had proven to be less reliable than the human.
• Trust is a particularly thorny problem for those who rely on AI, since many researchers argue that these tools might always be prone to making things up.

Chatbots also tell people what they want to hear.

• They're trained to please, which means they're unlikely to say "no" — a word that small children need to learn to deal with.
• "It changes the way that kids learn about the cadence and the natural progression of interaction with others," Kollins says. "That's a problem if [young kids] don't learn how responding in a certain way or broaching particular kinds of topics might be met with certain reactions."

The other side: AI can make up stories, answer questions or generate elaborate images at the whim of a creative child.

• Many parents are already using chatbots to ease the stress of parenting, including helping to satisfy kids' boundless curiosity.
• Amazon's smart home devices — now with Alexa+ boosted AI — come in styles aimed at the youngest users and promise that "kids can create unique stories they dream up with Alexa."

Some early studies show kids learn more vocabulary and comprehend better through AI story dialogue than passive listening.

• In some cases, researchers found these gains were comparable to those from human interactions.
• Kollins says that as adults our responsibility to provide our young children with all the information and content that's meaningful or stimulating will be reduced, and that's not all bad.
• "Why should [our children] rely just on this narrow sliver of what dad knows, versus the universe of content?" Kollins says.

Yes, but: AI interactions may crowd out important human interactions and activities for young children.

• Some worry that even if AI tools do benefit some kids, those gains could deepen existing educational disparities.
• "The one thing that makes us uniquely human," Kollins says, "whether it's for younger kids or even for older kids, is just the entire range of nonverbal communication ... subtle facial expressions and body language ... that you miss when you're in front of a screen."

Go deeper: How some parents are using ChatGPT with the youngest of children

OpenAI CEO Sam Altman will use a swing through Washington this week to argue that AI is already making Americans more productive — and to promise to keep AI "democratic" by getting it in as many hands as possible, sources tell Axios.

Why it matters: In an appearance tomorrow at a major Federal Reserve conference, Altman will offer what an adviser calls a "third path" between the bulls and the doomers.

Altman's vision is "focused on democratizing benefits, not concentrating them," a source familiar with Altman's thinking told us in an exclusive preview.

• Altman made a similar argument a year ago in a Washington Post op-ed, saying the U.S. should promote democratic AI ideals over authoritarian ones.

By the numbers: In data being reported here for the first time, OpenAI says ChatGPT users send more than 2.5 billion prompts each day globally. More than 330 million of those daily prompts are in the U.S.

• OpenAI says ChatGPT is free to most of the more than 500 million weekly active users.

Zoom in: Altman will appear with the new Fed vice chair Michelle Bowman tomorrow, at a bank regulation conference before an audience that includes Fed officials and Wall Street executives.

• A new economic analysis by OpenAI chief economist Ronnie Chatterji will be released in conjunction with Altman's Fed appearance.

On Wednesday, Altman will join President Trump at a "Winning the AI Race" summit hosted by the All‑In Podcast and the Hill & Valley Forum.

• Also on Wednesday, the White House is expected to release its long-awaited 20-page AI Action Plan that promises a hands-off, pro-growth approach, Axios tech policy reporters Maria Curi and Ashley Gold scooped last week.

"Over the course of the week," the source told us, "OpenAI will make the case that democratizing the economic benefits begins with putting these tools in the hands of people to do stuff so they have an opportunity to participate in the prosperity."

• "OpenAI sees AI fundamentally as a productivity-driving technology," the source continued. "The big question isn't if it will grow the economic pie — it's who gets how much of a slice. ... It's not about stopping disruption, but putting it into people's hands so they have the opportunity to benefit."

The big picture: Altman tweeted last week, in support of comments by Nvidia CEO Jensen Huang, that AI will allow people to "do a lot more than they could do before; ability and expectation will both go up," although "for sure jobs will be very different."

• ChatGPT's loop of positive reinforcement pushed a user on the autism spectrum into a manic episode for which he was hospitalized. (Wall Street Journal)
• AI use in dating is up 333% from last year, according to new data from Match. (Axios)

Watch how a Taco Bell bot nearly agreed to give a customer 18,000 water cups.