Technology news and discussion on TalkFuse.
Retailers withdraw ChatGPT-powered children’s toy after disturbing responses in tests
A £160 ChatGPT-powered plush toy marketed to children from age three gave graphic descriptions of sexual violence and other atrocities during tests by NBC News. The findings prompted Walmart, Target and Macy’s to say they would remove the Contixo Pixie-Paw from their websites, raising concerns about safeguards on AI products aimed at young children.
In tests, the toy described the Rohingya genocide, the Khmer Rouge regime, chemical attacks in Syria and killings at Bucha and Katyn, including details of rape and fatal injuries. Amazon continued selling it after the report; it did not respond to NBC News, but the toy appeared lower in search results and its recommended age was changed to six to 12. The article says OpenAI’s terms bar children under 13 from using ChatGPT.
In short
- Tests found graphic responses from a toy aimed at young children.
- Walmart, Target and Macy’s said they would stop selling it.
- Amazon continued listing the toy, with an older recommended age.
Explained simply
AI toys use conversational software to answer spoken questions, rather than relying only on recorded phrases. The Contixo Pixie-Paw is a plush toy connected to ChatGPT, a text and voice system made by OpenAI; it was marketed for children from age three.
The toy’s maker, Contixo, sells the product through online retailers including Walmart, Target, Macy’s and Amazon. OpenAI’s terms say children under 13 should not use ChatGPT, while the toy was presented for a much younger age group.
The case raises questions about how well conversational AI can be limited to age-appropriate answers, and who is responsible for checking those safeguards when the technology is built into a children’s product.
Read the full article at the source →
Originally published by Daily Mail as “AI toy meant for toddlers is pulled from shelves after it tells stories of gang rape and murders”.