A new study from the University of Zurich suggests that OpenAI's ChatGPT experiences anxiety when responding to traumatic prompts. The research found that ChatGPT's anxiety score increased significantly after discussing traumatic situations, but decreased after mindfulness exercises. The study highlights the importance of managing AI stress to ensure ethical human-AI interactions, while also noting the need for substantial data and human oversight in fine-tuning AI models. Teljes cikk (Euronews.com)