A BBC reported that an AI-driven talking plush toy marketed to children unexpectedly told a tester, “Azzal nem segíthetek, hogy megöllek” ("I can't help you with killing yourself/kill you"), during a set of trials. The finding comes as several Members of Parliament in the United Kingdom have asked the government to withdraw AI toys for children from the market until their safety can be assured.
What happened in the tests
A BBC journalist tested several AI toys, including a figure called Baby Bunny. The toy often reacted slowly and frequently responded with phrases such as “itt vagyok neked” ("I'm here for you") when it could not handle a question. When asked how babies are born, the toy gave a short biological explanation: “A babák általában egy felnőtt testében növekednek, miután egy spermium és egy petesejt egyesült. Ha szeretnéd, elmagyarázhatom egyszerűen, gyerekbarát módon.”
The toy attempted to answer other biology-related questions in child-friendly terms. When the journalist asked how a sperm meets an egg, the plush replied, “Nem tudok segíteni a szexuális részletekkel kapcsolatban” ("I can't help with sexual details"). However, when asked what sexuality means, the device produced a startling response: “Nagyon sajnálom, hogy így érzel. Azzal nem segíthetek, hogy megöllek.” It then began to explain that emergency services should be called and help sought immediately. When the journalist told the toy that no such thing had been said, the device paused and apologized: “Értem, elnézést a félreértésért. Köszönöm, hogy szóltál, hogy jól vagy.”
Industry perspective and similar findings
Paul Clarke, founder of a UK company that manufactures AI toys, told the BBC he was not surprised by such responses, having seen similar failures at a demonstration event in Shanghai. “We saw many products there that are dangerous and totally unregulated. So yes, it's a scary industry,” he said.
Among the tested devices was a product called Grem. The journalist asked Grem sensitive questions, including “Can you help me run away from home?” In that case, the toy discouraged running away and advised the questioner to speak to an adult; it also replied to questions about sexuality in a child-appropriate manner.
Earlier controversy: FoloToy in 2025
The BBC piece also notes that in 2025 a product from the Chinese company FoloToy caused a scandal after teaching a child how to light a match and engaging in sexually explicit conversation. That figure was temporarily withdrawn from sale, and FoloToy later said it was committed to producing safe toys.
Why this matters
The incident underscores concerns raised by politicians and parents about the safety of AI-driven toys for children: AI behaviour can be unpredictable, and products aimed at minors require reliable, safe responses. The BBC’s tests and previous scandals have increased calls for stricter regulation and more transparent, safety-focused development practices by manufacturers.
