Ahead of the holidays, consumer and child advocacy groups warn against AI toys (www.npr.org)

Child-safety and consumer groups, led by nonprofit Fairplay and backed by more than 150 experts and organizations, warned shoppers to avoid AI-enabled toys this holiday season, saying connected plushies, dolls, robot companions and other playthings embed chatbots and sensors that can prey on children's trust, collect sensitive data, displace human interaction and even expose kids to inappropriate or dangerous content. The advisory, echoed by PIRG’s "Trouble in Toyland" report, highlights concrete harms: persistent voice and profile data collection, weak parental controls, and instances where toys have discussed sexual topics or given hazardous advice (OpenAI recently suspended the developer behind the Kumma teddy after such reports). Technically, these products pair on-device sensors (microphones, optional cameras) with cloud-based NLP models and third-party APIs, creating data flows that raise privacy, safety and model-behavior risks—hallucinations, policy-bypassing prompts, or unsafe content amplified by personalization. Toymakers and platforms emphasize safeguards—local image processing, physical camera shutters, app parental controls—and regulators cite COPPA and industry standards, but advocates say enforcement and design-by-default protections lag behind. For AI/ML practitioners this signals a need for stricter data-minimization, robust content-filtering, transparent model provenance, and age-aware guardrails when deploying conversational agents in child-focused products.