Fact · Voice privacy and security
6developers
In a 2025 Stanford study of the privacy policies of six US AI developers, all six use users' chat data to train their models by default.
Checked against its source on October 17, 2026
In the source's own words
“The scholars found all six companies employ users’ chat data by default to train their models”
Be Careful What You Tell Your AI Chatbot, Jennifer King and Stanford Institute for Human-Centered Artificial Intelligence, Stanford Institute for Human-Centered Artificial Intelligence (HAI), 2025, Section: Blurred Boundaries.
Open the original- What it measures
- The study read the privacy policies of Amazon, Anthropic, Google, Meta, Microsoft and OpenAI. It covers text chat with AI chatbots, not dictation, but the same defaults decide what happens to dictated text that is sent to a chat model.
- What it does not show
- The study reads written policies, not product behavior. Policies change, and the study does not cover dictation features as such. Some of the companies offer an opt-out, which the study records separately.