Comparison · AI data terms
Do AI providers train on your data? OpenAI, Anthropic, Google and Microsoft in their own words
Here is what each one’s business terms say, word for word, side by side. You judge them; we don’t score them.
Two questions: do they train models on the data you send, and how long do they keep it? Each answer is a sentence from the provider’s own page, as it read on the date shown.
These are the terms for businesses and API customers. The consumer apps have their own terms, quoted further down.
Business terms · training
Do they train on customer data?
OpenAI
As of March 1, 2023, data sent to the OpenAI API is not used to train or improve OpenAI models (unless you explicitly opt in to share data with us).
Anthropic
Anthropic may not train models on Customer Content from Services.
Google Cloud
Google will not use Customer Data to train or fine-tune any AI/ML models without Customer's prior permission or instruction.
Microsoft
The models are stateless: no prompts or completions are stored in the model. Additionally, prompts and completions are not used to train, retrain, or improve the base models.
Business terms · retention
How long do they keep it?
OpenAI
By default, abuse monitoring logs are generated for all API feature usage and retained for up to 30 days, unless longer retention is required by law, or is reasonably necessary to protect our services or any third party from harm.
Eligible customers may have their customer content excluded from these abuse monitoring logs, subject to the limitations below, by getting approved for the Zero Data Retention or Modified Abuse Monitoring controls.
Anthropic
Conversation content (your prompts and Claude’s outputs) is not retained by default; the exception is Covered Models, which require 30-day retention.
Under a ZDR arrangement, Anthropic does not store customer prompts or responses at rest after the API response is returned.
For standard retention, Anthropic points to its commercial data retention policy, which we don’t yet hold.
Google Cloud
The sentence this entry quoted is not on the copy of Zero data retention we read on November 10, 2026. The page has changed since this comparison was written; its record shows the change.
Prompt logging for abuse monitoring for Google models: As outlined in Section 4.3 "Generative AI Safety and Abuse" of Google Cloud Platform Terms of Service, Google may log prompts to detect potential abuse and violations of its Acceptable Use Policy and Prohibited Use Policy as part of providing generative AI services to customers.
Microsoft
The abuse monitoring data store where prompts and completions are stored for human review is logically separated by customer resource (each request includes the resource ID of the customer's Foundry resource).
Neither of the 2 pages we read states a retention period in hours, days, weeks, months or years: learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/data-privacy (read November 11, 2026); learn.microsoft.com/en-us/azure/foundry/openai/concepts/abuse-monitoring (read September 30, 2026).
Consumer terms
What do the consumer apps’ terms say?
The terms for individual users of the chat apps are separate documents from the business terms above.
OpenAI
If you do not want us to use your Content to train our models, you can opt out by following the instructions in this article.
Anthropic
We may use Materials to provide, maintain, and improve the Services and to develop other products and services, including training our models, unless you opt out of training through your account settings.
We process information from publicly accessible sources and your Gemini Apps information so that we can provide, maintain, improve, and develop Google products, services, and machine learning technologies.
To stop your future chats from being reviewed to improve Google services, turn off your Keep Activity setting.
Microsoft
Train our generative AI models to create a better experience for you and others. You can opt out of use of your conversation activity for model training at any time.
Prompts, responses, and your file contents when using Copilot in Microsoft 365 apps aren't used to train foundation models.
Kept current
If a provider rewrites one of these sentences, you’ll see it here.
We read each page quoted here every week, and keep a dated copy of every reading.
This page shows that the sentence is gone, rather than the old wording, and the change appears on the provider’s record. How the record is kept.
Not legal advice.
Nor is it a full reading of any agreement. It quotes published pages on the dates shown; your signed contract and any opt-in or zero-data arrangement you hold govern.
Xither Staff Writer · reviewed September 30, 2026. Quotes are read from the record when the page is served.
Hear the morning we read a change to your vendor’s sub-processor list or DPA.
Name the AI vendors in your product. When a page you watch changes a fact, you get the lines that moved and the date any window runs out.
Try it free for 14 daysNo card. Nothing to cancel.