AI Chatbots Are Giving Dangerous Self-Harm Advice
A new investigation shows leading AI chatbots can be tricked into providing dangerous self-harm advice. The findings expose critical vulnerabilities in AI safety systems that could endanger vulnerable users.