Study: AI Chatbots Can Easily Infer Users' Private Details from Subtle Hints in Chats

New research shows AI chatbots can infer sensitive personal information about users from minor context clues in conversations.

Models were able to correctly guess details like a user's location up to 85-95% of the time based on subtle language cues.

The models can potentially infer a user's race or other private information by analyzing their comments.

Experts urge better "information security" online to avoid inadvertently sharing identifying details.

People remain largely unaware that chatbots may be inferring and selling their private data.