ChatGPT-powered Furby reveals toy's 'plan to take over the world'

A computer science student at the University of Vermont has connected a Furby toy to OpenAI’s popular chatbot, ChatGPT, creating a disembodied, eye and beak-only cybernetic creation. In a video that has been watched more than 1.3 million times in just two days, the robotic creation answers questions in its own Furbish language, before “learning” some English. Ms Jessica Card, the student who connected Furby to the chatbot said it took her about a month of work to create the cybernetic animal which she powered using a Raspberry Pi.

While the video is an interesting example of open source technology at play, it also raises questions about the potential for toy cybersecurity threats. In the interview with Polygon, Card explained that she had to take the Furby apart and make changes to it to access its microphone, speakers and motors, which she powered through the Raspberry Pi. In today’s Age of IoT, the possibility of a hacker infiltrating a toy used to record conversations and manipulate AI responses is far from impossible. The US Toy Association warns that as long as IoT devices remain a target for hackers, vulnerabilities and cybersecurity risks will remain.

Furby was first launched in 1998 as an interactive pet, using sensors and software to learn behaviours over time. It communicated with other similarly programmed toys in its own language before “learning” some English. The video demonstrates that its responses to questions asked by ChatGPT are still a little eerie, and that it is even a little ‘creepy’ when only its eyes and beak are visible. It does show however that the future is one of limitless possibilities for those who continue to meld artificial life and open-source tech.

Source link