AI Toy vs Smart Speaker: Classification Determines Test Scope
Classification Is a Procurement Decision, Not an Engineering One
The distinction between an AI toy and a smart speaker is not defined by microphone count, speaker wattage or wake-word latency. It is defined by how you position the device in the market — specifically, the age grade printed on the packaging and the language used in marketing materials. If the box says “Ages 4–8”, the lab tests for toy-specific hazards under EN 71-1/-2/-3 and EN 62115. If it says “For family use” with no age restriction, it enters the RED and EMC framework. This decision must be locked before tooling, because the test scope determines which standards apply, which lab gates open, and which documentation you must supply to customs. You can review our full how we build and test process to see how each hardware revision maps to certified test scopes.Age Grading Fixes the Regulatory Path — Before Any Unit Ships
Age grading is not advisory. Under Directive 2009/48/EC, it is legally binding and triggers mandatory conformity assessment. A device labelled “3+” must pass drop testing per EN 71-1, flammability per EN 71-2, and heavy-metal migration per EN 71-3 — even if its physical construction is identical to a non-toy model. That grading also determines whether the device qualifies for the “toy exception” in RED (exempt from certain SAR measurements) or remains subject to full radio emissions testing. Our E1 AI Learning Robot, for example, carries a “6+” grade and therefore undergoes full toy safety testing alongside its RED scope. You’ll find more guidance in our how to vet a supplier resource.Voice Functionality Alone Does Not Define the Category
A voice interface does not automatically make a device a smart speaker — nor does it disqualify it from being a toy. What matters is the primary function declared in the user manual and promotional copy. If voice is used solely for interactive storytelling, quiz delivery or phonics practice — and the device is marketed to children with age-graded packaging — it remains a toy. But if voice is presented as a general-purpose assistant (e.g., “Ask anything”, “Control your smart home”), it shifts into the RED domain. Our K1 AI Companion Robot is designed for multi-age interaction but avoids toy classification by omitting age grading and positioning voice as ambient assistance rather than play-based instruction.What This Process Cannot Do
This classification system cannot retroactively convert a smart speaker into a compliant toy — or vice versa — after production begins. If you ship units with “For ages 5+” on the box but have only booked RED and EMC tests, you have no valid CE marking for the toy market. There is no single “hybrid” test file accepted by EU notified bodies. Nor does any report exist that certifies both toy and audio equipment compliance in one go. Importers must reject suppliers who claim otherwise. Labs require separate applications, separate samples, and separate technical documentation packages — and WingPhoenix books each scope per hardware revision with accredited partners, as detailed in our factory testing page.Firmware and Content Are Part of the Classification Boundary
Content curation and firmware behaviour reinforce the declared category. A device sold as a toy must restrict access to adult-oriented content, disable open web search, and avoid third-party skill stores unless all skills are pre-vetted for child safety. Conversely, a smart speaker must support open voice assistants and interoperability with broader ecosystems. Our AI Learning & Robotics line uses closed-loop firmware with curated educational content — a design choice aligned with toy-grade responsibility, not a technical limitation. For customisation options including firmware white-label, refer to our customisation tiers.| Decision Point | Toy Path (EN 71) | Smart Speaker Path (RED/EMC) |
|---|---|---|
| Packaging label | Mandatory age grade (e.g., “3+”) | No age grade; “For home use” or similar |
| User manual | Child-focused safety warnings, play instructions | General setup, network configuration, privacy notice |
| Firmware features | No open web search, no unvetted skills, no adult content | Open assistant integration, third-party skill store, broad query handling |
| Lab test scope | EN 71-1/-2/-3, EN 62115, Toy Safety Directive | EN 300 328, EN 301 489, EN 62368-1 |
| Battery reporting | UN38.3 + SDS required; watt-hour marked on unit | UN38.3 + SDS required; watt-hour marked on unit |
| Importer obligations | WEEE, BattG, DEEE/Triman, EORI/VAT | WEEE, BattG, DEEE/Triman, EORI/VAT |
| Sample lead time | 7–15 days for configured sample | 7–15 days for configured sample |
| MOQ for custom firmware | 2,000 units | 2,000 units |
- Review your intended packaging, manual draft and firmware feature list before placing the first order
- Book toy or RED lab scope at the same time as firmware finalisation — delays here cascade into certification timelines
- Confirm importer responsibilities for WEEE, BattG and VAT registration before shipment
- Use our MOQ, sampling and lead times guide to align internal procurement cycles with factory capacity
- Refer to the full product catalogue to compare models already certified under each path
- Consult our the journal for quarterly updates on lab scope availability and regional policy shifts
Questions buyers ask
Why does the same device need two different test files?
Because EU and UK law classify by intended use and user age — not hardware. A voice-enabled device sold as a “3+ toy” triggers EN 71 and Toy Safety Directive; sold as a “smart speaker for home use” falls under RED and EMC directives instead.
Can I reuse test reports between categories?
No. Even identical PCBs require separate lab bookings: one for toy safety (mechanical, flammability, chemical migration), another for radio and EMC. Lab scope is tied to the declared product category on the test application.
Related: more on ai toys & dolls · buyer guides · customisation tiers