AI Chatbots Leak Real Phone Numbers to Users

A growing privacy concern has emerged as reports surface that AI chatbots, including Google's AI systems, are inadvertently exposing real phone numbers of individuals. Users have taken to social media and forums to share alarming experiences where their personal contact information was surfaced by AI models during routine queries. One Reddit user described a nightmare scenario: after an AI chatbot displayed their phone number in a response, they were flooded with calls from strangers seeking legal advice. The user, who is not a lawyer, had no idea how the AI obtained their number or why it was shared. This incident is not isolated; multiple users have reported similar breaches. The root cause appears to be the vast datasets used to train these models. AI systems are trained on publicly available information, including social media profiles, business directories, and public records. When a user asks a chatbot for contact details of a specific person or business, the model may retrieve and display phone numbers without proper verification or consent. What makes this issue particularly troubling is the lack of an easy opt-out mechanism. Unlike traditional data brokers, AI chatbots do not have a centralized system where individuals can request removal of their information. Once a number is in the training data, it can be surfaced repeatedly. Privacy advocates are calling for immediate regulatory action. They argue that AI companies must implement stricter filters to prevent the disclosure of personal contact information. Some suggest that chatbots should be programmed to refuse requests for specific phone numbers unless the user can prove authorization. Google and other AI providers have acknowledged the issue but have not yet announced comprehensive solutions. In the meantime, users are advised to be cautious about what information they ask AI assistants to retrieve. This incident serves as a stark reminder that while AI offers incredible convenience, it also poses serious risks to personal privacy that have yet to be fully addressed.