Do AI companies sell your personal data?

Many people worry that chatbot companies sell what they type into them. The honest answer is more specific than yes or no: the four biggest AI providers say they do not sell personal data in the legal sense, but they can use consumer conversations to train their models unless you opt out.

This guide sets out what OpenAI, Anthropic, Google and Microsoft say in their own privacy policies and help pages, as checked in November 2026, and how to change the settings. Policies change often, so check the linked pages before relying on them.

Do AI companies sell your personal data?

AI companies OpenAI, Anthropic and Microsoft say in their own policies that they do not sell personal data as US state privacy laws define a sale. Google's policy does not use the word sell, but says it does not share personal information outside Google except in listed cases.

A sale under laws such as the California Consumer Privacy Act is a disclosure of personal information for money or other value. Sharing data for targeted advertising is regulated separately, which is why Microsoft offers an opt-out from sharing and says it honors the Global Privacy Control browser signal.

Privacy promises also limit later changes. FTC staff wrote in February 2024 that a company adopting more permissive data practices, such as AI training, through a quiet, retroactive change to its terms or privacy policy could be acting unfairly or deceptively. That is why the providers above ask users to choose, or announce changes in advance, as Anthropic did in August 2025.

Do AI companies train their models on your chats?

AI companies can train their models on your consumer chats. OpenAI and Microsoft do so unless you opt out, Google does so while Keep Activity is on, and Anthropic does so only if you allow it. Training is a use of your data inside the company, not a sale, so a no-sale promise does not stop it.

OpenAI

OpenAI's help article on how your data is used to improve model performance says that after you opt out, new conversations will not be used for training. Its privacy policy says it does not apply to content processed for customers of its business offerings.

Anthropic

Anthropic announced updates to its consumer terms on August 28, 2025. Users who allow their chats to be used for model improvement have data retained for five years, and others stay on the existing 30-day retention period. The Anthropic Privacy Policy adds that if you rate an output, Anthropic may store the entire related conversation as feedback.

Google

Google's Gemini Apps Privacy Hub says "A subset of chats are reviewed by human reviewers" and that reviewed chats are kept for up to three years, disconnected from your account, even if you delete your activity. With Keep Activity off, Google still keeps chats for 72 hours to respond and to protect its users. The page advises against entering confidential information you would not want a reviewer to see.

Microsoft

The Microsoft Privacy Statement, last updated September 2026, says "we may use your data to develop and train our AI models". Microsoft's Copilot privacy FAQ for the older app says it does not train on users signed in with an organizational Entra ID account, signed-in users under 18 or users in Brazil, China excluding Hong Kong, Israel, Nigeria, South Korea and Vietnam. Microsoft says an updated Copilot app has been available since August 18, 2026, with its own privacy article, so check the settings in the version you use.

You cannot sell your personal data to AI companies through Sell My Data To AI. We work only with companies licensing records they own, and every seller passes our eight-point vetting standard, which covers authority to sign, ownership, privacy notices and de-identification.

The data AI buyers want from companies is mostly business records with personal details removed, such as code, support tickets and documents. Our guide to what data AI companies buy explains the demand. If your employer gives you ChatGPT Enterprise, Claude for Work or Microsoft 365 Copilot, your work chats fall under the company's business terms, not the consumer settings above. If you run a company with records like these, request a free data valuation.

What privacy rights can you use over your data?

You can use legal privacy rights to see, delete or stop the sale or sharing of your data, and to object to its use for AI training in some countries. Which rights you have depends on where you live.

The right to object has been used for AI training in Europe. Ireland's Data Protection Commission said on May 21, 2025 that Meta had introduced a simpler in-app objection form before training on adults' public posts. Our data rights page explains how to make a request, and our page on AI training data lawsuits tracks the court cases.

How do you stop AI companies using your data?

You stop AI companies using your data mainly through their settings, then through legal requests where settings are not enough.

1. 1

   Change each app's training setting

   Use the opt-outs in the training table above for every AI app you use, and repeat it on each account.

2. 2

   Use temporary or private modes

   Use Temporary Chats in Gemini, and avoid entering confidential information that a human reviewer could see.

3. 3

   Turn off ad sharing

   Use each company's advertising opt-out, or turn on the Global Privacy Control signal in your browser.

4. 4

   Make a privacy request

   Ask for access or deletion under your state law or the GDPR, using the company's privacy request page.

5. 5

   Clear data broker records

   If you live in California, submit one request through DROP to reach registered data brokers.

These steps limit future use. Data already used in training may not be removable from a trained model, so the settings matter most before you share anything. Our guide to de-identification explains how companies that license data are expected to strip personal details first.