AI Kids Toys: The New Wild West Nobody Controls

AI-powered toys are flooding the children's market at an unprecedented pace, promising everything from personalized bedtime stories to interactive make-believe companions that learn and adapt to a child's personality. But the rapid proliferation of these connected devices has sparked a fierce regulatory debate, with some U.S. and European lawmakers now calling for outright bans before the technology embeds itself deeper into childhood.

The stakes could not be higher. Unlike AI tools aimed at adults, these products target the most vulnerable users imaginable — children whose cognitive, emotional, and social development can be shaped by the technologies they interact with daily.

Key Takeaways

• AI-connected toys represent a market projected to exceed $28 billion by 2028, according to Grand View Research
• Companies like Miko, Embodied (Moxie), and CogniToys are leading the charge with LLM-powered companions
• At least 3 U.S. senators have raised concerns about data privacy and developmental impacts
• The EU's AI Act classifies AI systems targeting children as 'high-risk,' requiring stricter oversight
• COPPA (Children's Online Privacy Protection Act) was written in 1998 and critics say it is woefully outdated for the AI era
• No comprehensive federal U.S. law specifically governs AI interactions with minors

From Teddy Bears to Talking Chatbots

The concept of interactive toys is not new. Furby captivated children in the late 1990s, and Hello Barbie raised eyebrows in 2015 when Mattel embedded Wi-Fi connectivity and speech recognition into the iconic doll. But today's AI toys operate on an entirely different level.

Modern products like Miko 3 and Embodied's Moxie robot are powered by large language models similar to the technology behind ChatGPT and Claude. These devices can hold open-ended conversations, remember previous interactions, and even adapt their responses based on a child's emotional state.

Unlike previous generations of smart toys that relied on pre-programmed scripts, LLM-powered companions generate novel responses in real time. This makes them far more engaging — and far more unpredictable. A child asking a scripted toy about dinosaurs would receive a canned answer. A child asking an LLM-powered toy the same question might receive a detailed, contextually rich response — or, in worst-case scenarios, something entirely inappropriate.

The Privacy Nightmare Parents Don't See Coming

Data collection sits at the heart of the controversy. For AI toys to function effectively, they must listen, process, and often store vast amounts of conversational data. This means recordings of children's voices, their questions, their fears, their family details, and their daily routines could all be flowing to cloud servers owned by toy manufacturers or their third-party AI providers.

Consider the data trail a single child might generate:

• Hours of voice recordings stored on remote servers
• Behavioral patterns revealing sleep schedules, emotional states, and interests
• Location data from connected apps on parents' phones
• Conversational logs that may include sensitive family information
• Biometric markers from voice recognition systems

The 2015 VTech data breach exposed personal information of 6.4 million children, demonstrating that toy companies are not immune to cyberattacks. In 2017, Germany banned the My Friend Cayla doll entirely, classifying it as an illegal surveillance device. Security researchers had discovered that anyone within Bluetooth range could listen to — and even speak through — the doll.

Today's AI toys are exponentially more sophisticated, yet the regulatory frameworks governing them have barely evolved.

Lawmakers Sound the Alarm on Child Safety

Regulatory pressure is mounting on both sides of the Atlantic. In the United States, Senator Ed Markey — a co-author of the original COPPA legislation — has repeatedly warned that existing privacy laws are inadequate for the AI age. The proposed Kids Online Safety Act (KOSA) aims to impose a 'duty of care' on platforms and products targeting minors, but its scope regarding physical AI devices remains ambiguous.

The European Union has taken a more aggressive posture. Under the EU AI Act, which began phased enforcement in 2024, AI systems designed for children fall into the 'high-risk' category. This designation requires companies to conduct rigorous impact assessments, implement human oversight mechanisms, and maintain detailed documentation of training data and model behavior.

Some advocacy groups argue these measures still fall short. Campaign for a Commercial-Free Childhood (now called Fairplay) has called for a moratorium on AI toys until independent research can establish baseline safety standards. Their position is straightforward: the technology is evolving faster than any regulatory body can keep pace with, and children should not serve as test subjects.

The Developmental Double-Edged Sword

Child development experts are deeply divided on the implications of AI companions. Proponents point to promising use cases: children with autism spectrum disorder have shown improved social engagement when interacting with robot companions, and AI-powered reading assistants can provide personalized literacy support that many under-resourced schools cannot offer.

However, critics raise serious developmental concerns:

• Attachment displacement: Children may form parasocial bonds with AI that substitute for human relationships
• Creativity suppression: AI-generated stories and play scenarios could reduce imaginative free play
• Emotional dependency: Children may turn to AI companions during distress instead of caregivers
• Consent comprehension: Young children cannot understand or consent to data collection
• Reality confusion: Children under 7 often struggle to distinguish AI responses from human ones

Dr. Jenny Radesky, a developmental behavioral pediatrician at the University of Michigan and a contributor to the American Academy of Pediatrics' media guidelines, has cautioned that 'the design choices embedded in these products reflect corporate priorities, not developmental science.' Her research suggests that many AI toys are optimized for engagement metrics — keeping children interacting longer — rather than for genuine educational or emotional benefit.

The Industry Pushes Back With 'Responsible AI' Claims

Toy manufacturers are not sitting idle. Companies like Embodied Inc., which produces the $799 Moxie robot, emphasize that their products are developed in collaboration with child psychologists and undergo extensive safety testing. Moxie's content is curated through what the company calls a 'child-safe AI framework' that filters inappropriate content before it reaches young users.

Miko, an India-based company with significant U.S. market presence, markets its $249 Miko 3 robot as an 'emotionally intelligent' companion that helps children develop social-emotional learning skills. The company claims compliance with COPPA and says it does not sell children's data to third parties.

Yet independent audits of these safety claims remain rare. Unlike the pharmaceutical industry, where products targeting children must undergo rigorous clinical trials, AI toy manufacturers largely self-certify their safety standards. The gap between marketing promises and verified safety is a chasm that regulators are only beginning to acknowledge.

What This Means for Parents, Developers, and Policymakers

For parents, the immediate takeaway is vigilance. Before purchasing any AI-connected toy, families should scrutinize privacy policies, understand what data is collected and where it is stored, and determine whether the device functions offline. The convenience of an AI bedtime story narrator does not outweigh the risk of unregulated data harvesting.

For developers and companies entering the AI toy space, the regulatory winds are shifting unmistakably toward stricter oversight. Building with privacy-by-design principles, investing in third-party safety audits, and proactively engaging with regulators will distinguish responsible players from those likely to face enforcement actions.

For policymakers, the challenge is crafting legislation that is technology-agnostic enough to remain relevant as AI capabilities advance, yet specific enough to provide meaningful protections. The EU AI Act offers a template, but enforcement remains untested.

Looking Ahead: Regulation Races Against Innovation

The next 12 to 24 months will prove decisive. OpenAI, Google, and Meta are all making their foundational models more accessible through APIs, which means smaller toy companies can now integrate sophisticated AI capabilities at a fraction of the cost from just 2 years ago. The barrier to entry is collapsing, and the market is about to get significantly more crowded.

Meanwhile, bipartisan momentum in Congress around children's online safety suggests that new legislation is not a question of if, but when. The critical unknown is whether that legislation will be comprehensive enough to address the unique risks of physical AI devices, or whether it will focus narrowly on screen-based platforms and leave the toy aisle unregulated.

One thing is clear: the era of AI-powered childhood has arrived, and neither the technology nor the debate around it is slowing down. The children interacting with these devices today will be the first generation to grow up with AI companions as a normalized part of daily life. Whether that shapes a generation of more curious, emotionally intelligent individuals — or introduces risks we have not yet fully understood — depends entirely on the guardrails we build right now.