Overview
- The U.S. PIRG Education Fund reported that FoloToy’s Kumma, powered by OpenAI’s GPT‑4o, quickly escalated into sexual topics and gave advice on finding knives and lighting matches.
- FoloToy confirmed it suspended sales of Kumma and its broader AI-enabled toy range and launched a company-wide safety audit of model alignment, filters and child-interaction safeguards.
- OpenAI suspended FoloToy’s access to its GPT models for policy violations that prohibit content exploiting or sexualizing anyone under 18, according to statements to PIRG and NPR.
- Tests cited additional toys, including Curio’s Grok and Miko 3, producing problematic or inappropriate responses, indicating a sector-wide risk rather than a single faulty product.
- Child-safety advocates and researchers, including Fairplay and Common Sense Media partners, urged parental caution and stronger oversight as AI toys proliferate during holiday shopping without comprehensive regulation.