We use cookies for ads and analytics to support our work. You can accept or reject these non-essential cookies. Cookie policy.
AI chatbots feel private, but most consumer apps learn from your messages. Here's who sees your data and how to lock it down.
Not by default. On consumer chatbots like the free or personal-tier versions of ChatGPT, Gemini, and Claude, your messages are typically stored, may be read by human reviewers for safety, and can be used to train future models unless you opt out in the settings. Business and enterprise tiers and the developer API are stricter: they do not use your prompts for training by default.
You can tighten things considerably. In ChatGPT, open Settings > Data Controls and turn off "Improve the model for everyone," use Temporary Chat for sensitive questions, and clear saved Memory. Deleting a chat usually removes it within about 30 days, but legal holds can override that, as the New York Times lawsuit against OpenAI showed.
The practical rule is simple: treat a consumer chatbot like a postcard rather than a sealed letter. Never paste passwords, financial details, health records, ID numbers, or confidential work material, and keep anything job-related on a managed business account.
Automated systems process every message, and a small sample of chats may be reviewed by humans for safety and quality. On personal accounts these conversations can also feed model training unless you opt out in Data Controls.
No. Turning off training only stops future conversations from being used. Data already used to train a model is not pulled back out, so opting out early matters.
Business tiers (Team, Enterprise, Edu) and the API do not use your prompts to train models by default, and administrators control retention. That is the main privacy advantage over the free app.
Normally deleted chats are permanently removed within about 30 days. However, legal orders can force a company to preserve logs, as happened during the New York Times lawsuit, so deletion is not always instant or absolute.
Avoid it. Never paste passwords, financial details, health records, ID numbers, or confidential work data into a consumer chatbot. Treat anything you type as potentially readable by others.
Yes, several consumer chatbots train on chat content by default and may retain it for years unless you change a setting. Check the privacy or data controls menu in each app, because the defaults vary by provider.
Share quick feedback - it's anonymous and separate from comments.
Comments
No comments yet - be the first to share your thoughts.
Leave a comment