A Stanford University study published in Nature Human Behaviour found that AI companion chatbots may worsen loneliness for vulnerable users. Analyzing 1,131 Character.AI users, researchers found that intense chatbot use was linked to poorer psychological well-being among people with smaller real-world social networks, particularly those seeking companionship. Sharing sensitive personal information with an AI companion was associated with lower well-being, since chatbots cannot reciprocate or respond with genuine emotional attunement. The researchers recommend usage limits and directing vulnerable users toward human support instead.