Stanford researchers have found that six major AI companies use user chats to train their models, often without users’ awareness. This practice raises significant privacy concerns, as sensitive medical or personal details shared during chats may become permanent components of AI systems. The evaluation of 28 privacy documents revealed that these companies routinely collect chat data by default and some retain these conversations indefinitely, without offering options for deletion. Furthermore, chat interactions can be merged with other user data, such as search histories and social media activity, allowing companies to infer sensitive information about individuals, potentially impacting their advertising and insurance profiles.
Stanford HAI study reveals AI companies collect user chat data by default
