AI toys are now squarely inside the EU's AI Act — and inside a parallel debate in the US Congress. Neither framework bans AI toys outright. Both draw a hard line around toys designed to manipulate children, and add new paperwork for everything else.
Does the EU AI Act ban AI toys?
No, not categorically. Under the EU AI Act, AI systems embedded in toys are classified as "high-risk" when they act as a safety component of the product, which brings extra documentation, testing, and human-oversight requirements. What is banned outright is narrower and stricter: any AI system, in a toy or otherwise, capable of exploiting a child's age-related vulnerabilities to materially distort their behavior in a way likely to cause harm falls under the Act's "unacceptable risk" category and is prohibited.
When do the high-risk rules for toys actually apply?
Later than many parents assume. Following the "Digital Omnibus on AI" agreed in May 2026 and approved by the Council on 29 June, the timeline for high-risk product rules — including toys — was pushed back to 2 August 2028, according to Skadden's 2026 AI Act update. The outright ban on manipulative AI systems is not delayed in the same way.
How does this compare with the US?
The US is still assembling its version. The Senate passed the Children and Teens' Online Privacy and Protection Act ("COPPA 2.0") in March 2026, and its provisions were folded into the broader KIDS Act, which the House passed in June 2026. As of this writing it has not been reconciled between chambers or signed into law, per Loeb & Loeb's 2026 analysis. In the meantime, the FTC's updated COPPA rule — enforceable since April 2026 — already requires clearer notice, consent, and data minimization from any connected toy collecting children's data.
| Question | EU (AI Act) | US (current law + pending bills) |
|---|---|---|
| Manipulative AI aimed at kids | Prohibited outright | Restricted under FTC COPPA rule; KIDS Act would expand this |
| High-risk toy AI requirements | Apply from August 2, 2028 | No federal high-risk toy category yet |
| Data use for advertising to kids | Heavily restricted | Restricted under COPPA; KIDS Act would extend to teens |
| Status as of August 2026 | Law, phased implementation | KIDS Act passed House only, not yet law |
What should this mean for how parents shop today?
Regulation is catching up to a category that grew faster than the rules around it. Until the 2028 deadline, most of what protects a child using an AI toy today is what the manufacturer chose to build in voluntarily — not what the law requires yet. That's worth checking directly: does the toy have a camera, does it require a subscription that changes terms later, and does the privacy policy rule out data resale. We cover what to look for in more detail in our explainer on toy safety certifications and our 2026 roundup of AI toy safety laws.
Ted was built around the strict end of these emerging rules rather than the minimum required today: no camera, no ad targeting, and no data resale, with full detail on our security and certifications page.
FAQ
Are AI toys illegal in the EU?
No. Most AI toys remain legal; they will need to meet high-risk AI system requirements from August 2028, and manipulative designs are already banned.
Is COPPA 2.0 in effect yet?
No. Its provisions were absorbed into the KIDS Act, which passed the House in June 2026 but has not been reconciled with the Senate or signed into law.
What does "manipulative AI" mean under the AI Act?
Systems designed to exploit a child's age-related vulnerabilities to distort their behavior in ways likely to cause harm — this category is banned outright, with no phase-in period.
Does GDPR still apply to AI toys in the meantime?
Yes. GDPR's rules on consent and data minimization for children's data apply regardless of the AI Act's toy-specific timeline.