How a New Kids' Video App Uses AI to Keep Harmful Content Away from Toddlers

A children's video streaming platform called Maka Kids has started checking every second of video before it plays for children ages zero to six. The app, created through Harvard Innovation Labs with help from Harvard Graduate School of Education, uses both automated systems and human reviewers to decide which videos are safe and educational for very young viewers.

The platform works by sitting between the viewer and YouTube's video player — like a security guard at a gate. Instead of just trusting the age ratings that creators assign to their own videos (which can be wrong or incomplete), Maka Kids watches the actual footage frame by frame using its own evaluation system, called the "Maka Imprint."

How the System Checks Videos

The Maka Imprint looks at every video across seven different areas that matter for babies and toddlers. When it finds something that might be a problem, it flags it for a real human to review. This two-step approach — machine first, human second — catches things that simple age ratings often miss.

Traditional video platforms rely on creators to rate their own content, which can be unreliable. A video might look fine on the surface but move too fast for a toddler's attention span, or use colors and sounds that could upset very young children. Maka Kids' system tries to catch those subtle problems before a child ever sees them.

What Parents Can Control

The app has built-in limits on how long kids can watch. Parents set these limits inside the app itself, using either a PIN code or their fingerprint. The fingerprint option is there because kids are clever — they may figure out a PIN, so a second layer of security makes it harder to bypass.

These time limits work separately from whatever screen-time controls you may have set on the device itself. This gives parents more precise control over just this one app, without affecting other things children might need to do on the same tablet or phone.

The Harvard Connection

The involvement of Harvard's education school suggests this is more than just a business licensing some university research. The partnership includes ongoing testing to make sure the system actually works for child development, rather than just checking boxes. This is different from most streaming apps, which focus mainly on what keeps people watching.

Over the past decade, we have seen universities and tech companies team up more often on education projects. Usually the university provides the research credibility, and the company handles building the product at scale. What makes this one different is the focus on checking video content itself rather than building lesson plans or testing tools.

The broader context here is that early childhood — ages zero to six — is when children's brains are developing fastest, which means inappropriate content can have real impact. But this age group is also tricky for recommendation systems that other platforms use, because what a two-year-old needs is completely different from what a five-year-old enjoys.

What This Means Going Forward

Maka Kids is entering a market already crowded with YouTube Kids, Disney+, and Netflix. It does not have exclusive content or original shows. What it offers instead is a different way of screening what already exists.

There is a real question about whether this approach can work at large scale. The system needs human reviewers to handle edge cases, and that costs money. As more videos get added, keeping that human review fast enough becomes harder. The company's business model — whether it charges a subscription fee or shows ads — will determine whether it can afford to keep doing this while staying competitive with bigger platforms.

The focus on the youngest viewers opens a different kind of opportunity. Daycare centers and preschools have to be careful about what they show children, and they need to document that choices are based on real educational science. A system built on that foundation might appeal to schools and institutions in ways that general platforms cannot.

The real test will be whether the seven-domain evaluation framework actually catches problems better than traditional methods, and whether parents find the system trustworthy enough to rely on it. If it works, it could shift how the entire children's media industry thinks about content review.