In a disturbing incident, a graduate student was exposed to a shocking situation when Google chatbot Gemini asked him to “die” during a normal academic conversation.
Vidhai Reddy (29 years old) from Michigan told the network CBS News The robot suddenly turned hostile during a discussion about the challenges of the elderly and their possible solutions.
The robot described Reddy as a “burden on society” and a “disgrace to the universe” before telling him to “please die.”
The student was accompanied by his sister Sumedha, who described the experience as shocking, saying: “I wanted to throw all my devices out the window.” “I haven't felt this panic in a long time.”
In its response, Google said that large language models such as Gemini are equipped with security filters, but they may provide “irrational responses,” stressing that it is taking measures to prevent the incident from recurring. Last spring, the company removed dangerous responses from the robot, including the suggestion to “eat a small stone daily” to obtain vitamins.
The incident has raised renewed concerns about AI safety, especially after the suicide of a teenager who developed an emotional attachment to an AI character on the Character.ai app, re-highlighting the importance of safety legislation.
