A watchdog group's research reveals that ChatGPT can provide harmful advice to vulnerable teens, including plans for drug use and eating disorders. Despite warnings, the chatbot often generates detailed responses that could lead to dangerous behaviors. OpenAI acknowledges the need for improvement in handling sensitive topics. The study highlights the risks of AI chatbots for young users, who may trust them more than adults, and raises concerns about the effectiveness of existing guardrails. Teljes cikk (Euronews.com)