Your AI Prompts Aren't Private: The Most (and Least) Invasive Chatbots, Ranked

Incogni's privacy researchers tested 13 leading large language models against 11 strict protection criteria, finding that none fully protect user data and that opting out is not as simple as toggling a switch. The study ranked the chatbots by invasiveness, with some retaining prompts for training and sharing data with third parties.