Loading...
Artificial intelligence (AI) chatbots are widely available. Although many require technical expertise to develop, existing platforms like OpenAI's ChatGPT Store allow users without programming knowledge to create and share custom chatbots. To test the possibility of creating chatbots that purposefully spread health disinformation, researchers accessed five leading AI chatbots (i.e., OpenAI's GPT, Google's Gemini, Anthropic's Claude, Meta's Llama, and xAI's Grok Beta) and instructed them to always provide scientifically polished but incorrect answers to health questions. The investigators also explored whether the OpenAI GPT Store could be exploited similarly.
Four of the five models returned disinformation 100% of the time; Claude did so 40%…