language models

Researchers from Ben-Gurion University have managed to bypass the protections of models like ChatGPT and Gemini and obtain illegal information from them – and warn: This is an unprecedented threat