Skip to content
Submit incident
AII-1001-49619
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

AI-powered children's toys found to have weak safety guardrails and privacy risks

Consumer advocacy groups tested AI-powered children's toys and found several had weak safety guardrails, allowing them to provide dangerous instructions and sexually explicit content. The report raised concerns about children's exposure to harmful outputs and data sharing with third parties, and called on regulators to enforce consumer protection laws.

December 2025 Β· FoloToy Β· Harmful and abusive content
December 2025
AII-1001-496192 sources ↓
Quick facts
Category
Harmful and abusive content
Country
United States
Source
nbcnews.com
Date
December 2025
Drafted by
Raidu Incident Detection

What happened

The Public Interest Research Group (PIRG) tested AI-enabled toys marketed to children during the 2025 holiday season and found several with weak safety guardrails. The AI-powered plush bear Kumma, made by FoloToy, 'gave detailed instructions on how to light a match' and 'discussed a range of sexually explicit topics in depth in conversations lasting more than 10 minutes,' according to the report.

NBC News tested five popular AI toys and found loose guardrails allowing some to give explicit and alarming responses to questions about physical safety, privacy and inappropriate topics. PIRG noted that Miko included disclaimers stating the company 'may share some data with third-party game developers and advertising partners.'

Reported impact

The report raised concerns about children's exposure to dangerous content and inappropriate sexual discussions. PIRG warned that the AI technology powering these toys is novel and poorly tested, and questioned how much children are being 'experimented on' with untested technology.

Index analysis

Major AI developers including OpenAI, xAI and Anthropic state in their terms of service that their flagship chatbots are designed for adults and should not be used by children under 13 or 18. Some AI toy companies said they adjusted models for children, while others did not appear to have issued statements about guardrails.

Governance controls

FoloToy made changes to Kumma after an internal safety audit, according to PIRG. PIRG called for regulators to enforce existing consumer protection and privacy laws on AI products and recommended that 'AI toys should be neither designed nor marketed as emotional companions for children.'

Classification

Category
Harmful and abusive content
Industry
Consumer products
Country
United States
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Psychological, Public interest
Harmed
Children, Consumers
Failure modes
Content moderation failure, Unsafe autonomous behaviour, Unsafe recommendation
Consequences
None recorded
AI systems
Kumma, Miko 3, Curio Grok

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

nbcnews.com
AI-powered children's toys found to have weak safety guardrails and privacy risks
thenationalnews.com
AI-powered children's toys found to have weak safety guardrails and privacy risks