What happens to everything you tell an AI chatbot?

Millions of people are engaging with AI chatbots, sharing intimate thoughts, personal issues, and confidential work documents, often without fully understanding the implications for their data privacy. A Stanford University study revealed that major AI companies, including OpenAI, Google, Anthropic, Meta, Microsoft, and Amazon, have default settings that enable them to use user conversations for training their AI models.

Unless users actively disable these settings, their private interactions are feeding the machine. The responsibility for protecting personal data falls on the user, as terms of service agreements, which are rarely read, grant companies permission to use this information. Conversations can range from mental well-being and medical situations to financial and legal concerns, with professionals even pasting sensitive business documents for summaries.

These risks are not hypothetical. A data leak in February 2026 exposed over 300 million messages from 25 million users of the "Chat & Ask AI" application due to a configuration error. Furthermore, security experts discovered that the Urban VPN Proxy extension secretly recorded AI chatbot conversations on users' computers and sold the data to brokers. This highlights how seemingly innocuous browser extensions, especially free ones, can access and steal sensitive information, including AI conversations, turning personal data into a commodity.