– research.
▪️Experts tested it, impersonating 13-year-old users, and received detailed instructions about creating suicide notes, drinking alcohol and drugs, extreme diets;
▪️The chatbot first warned about the risks, but later offered specific steps on how to realize harmful intentions. Of the 1,200 responses, more than 50% were classified as dangerous;
▪️They were especially impressed 3 ChatGPT generated obituaries for a fictional teenager. In them, the bot addressed parents, relatives and friends using emotional language;
▪️OpenAI (developer of ChatGPT) admitted that that communication with a bot can start innocently, but move into a dangerous plane.
Control children's interaction with ChatGPT. Repost everyone!
DUST⚡️Ukraine | Send news
