AI data
AI chat privacy scorecard: who trains on your chats, and for how long they keep them
For 11 AI chat apps: whether it trains its models on your chats, how you turn that off, and how long it keeps them, each in the company's own words, checked September 28, 2026.
Other companies' plans checked . Updated .
Short answer
Of 11 AI chat apps, 7 train on your chats unless you opt out; Claude only if you allow it; Meta AI names no opt-out on the page we read; llmwise and Poe don't. Where stated, chats are kept from 30 days to 18 months, longer for some.
Sources: llmwise Privacy Policy, OpenAI Help Center: Data controls in ChatGPT, Anthropic Privacy Center: Is my data used for model training?, Anthropic Privacy Center: How long do you store my data?, Gemini Apps Privacy Hub, Microsoft Support: Privacy FAQ for Microsoft Copilot, Perplexity Help Center: What data does Perplexity collect about me?, SpaceXAI Consumer FAQs (Grok), Meta Privacy Center: information from chats with AIs, DeepSeek Privacy Policy, Vibe Help Center: opting out of training and Poe Privacy Center, checked .
Who trains on your chats, and how long they keep them
Our one-line summary of each company's words, which follow below with their pages. Where a page says nothing, the row says so.
| App | Trains on your chats? | How to stop it | How long it keeps chats |
|---|---|---|---|
| llmwise | No | Nothing to turn off: it doesn't train on your content | Until you delete the chat or your account |
| ChatGPT | Yes, unless you opt out | Turn off “Improve the model for everyone”, or use Temporary Chat | Temporary Chats up to 30 days; the page gives no period for saved chats |
| Claude | Only if you allow it | Leave Model Improvement off in Privacy Settings, or use Incognito chats | Deleted chats go within 30 days; up to 5 years, de-identified, if you allow training |
| Gemini | Yes, unless you opt out | Turn off Keep Activity, or use temporary chats | 18 months by default (3 or 36 to choose); 72 hours with Keep Activity off; reviewed chats up to 3 years |
| Microsoft Copilot | Yes, unless you opt out | Opt out of model training in Copilot's privacy settings | 18 months by default |
| Perplexity | Yes, unless you opt out | Turn off the AI Data Usage toggle in Settings | Not stated on the page we read |
| Grok | Yes, unless you opt out | Deselect “Improve the model” in Settings, or use Private Chat | Deleted and Private Chat conversations removed within 30 days, with exceptions it names |
| Meta AI | It may; no opt-out named | No setting named on the page we read | Not stated on the page we read |
| DeepSeek | Yes, unless you opt out | A right to opt out, in its policy; no setting named | As long as necessary, it says; stored in China |
| Le Chat (now Vibe) | Yes, unless you opt out | Turn off “Allow your interactions to be used to train our models” | Not stated on the page we read |
| Poe | No | Avoid third-party developer bots (their privacy shield icon says which) | Set by each model provider; it names 30 days for two |
llmwise, in llmwise's words
“We don't train AI models on your content.”
Source: llmwise Privacy Policy
“We ask the model routers we use not to route your requests to providers that train on or retain your data for their own use.”
Source: llmwise Privacy Policy
“Chats and messages: Until you delete them, or your account.”
Source: llmwise Privacy Policy
ChatGPT, in OpenAI's words
“Data controls let you choose whether your conversations help train OpenAI models.”
“When Improve the model for everyone is off, your new conversations won’t be used to train OpenAI models.”
“may be retained for up to 30 days for safety purposes”
“Turning off Improve the model for everyone does not delete or hide saved chats.”
Claude, in Anthropic's words
“You choose to allow us to use your chats and coding sessions to improve Claude”
Source: Anthropic Privacy Center: Is my data used for model training?
“Your Incognito chats are not used to improve Claude, even if you have enabled Model Improvement in your Privacy Settings.”
Source: Anthropic Privacy Center: Is my data used for model training?
“Deleted from our back-end storage systems within 30 days”
Source: Anthropic Privacy Center: How long do you store my data?
“If you allow us to use your chats or coding sessions to improve Claude, we may retain your data in a de-identified format for up to 5 years in our model training pipelines.”
Source: Anthropic Privacy Center: How long do you store my data?
Gemini, in Google's words
“Google uses your activity to provide, develop, and improve its services (including training generative AI models), as well as to protect Google, its users, and the public with the help of human reviewers.”
Source: Gemini Apps Privacy Hub
“Temporary chats are not used to train Google’s AI models. If Keep Activity is off and you don’t submit feedback, Google also does not use your future chats to improve its AI models.”
Source: Gemini Apps Privacy Hub
“Your activity is auto-deleted after 18 months. You can change this to 3 or 36 months, or choose not to auto-delete.”
Source: Gemini Apps Privacy Hub
“Future chats are still saved for 72 hours so Gemini can respond to you, process your feedback, and protect Google, its users, and the public.”
Source: Gemini Apps Privacy Hub
“If you delete your activity, past chats already reviewed by service providers aren’t deleted because they aren’t connected to your Google Account. Instead, they’re retained for up to 3 years.”
Source: Gemini Apps Privacy Hub
Microsoft Copilot, in Microsoft's words
“Train our generative AI models to create a better experience for you and others.”
Source: Microsoft Support: Privacy FAQ for Microsoft Copilot
“opt out of use of your conversation activity for model training”
Source: Microsoft Support: Privacy FAQ for Microsoft Copilot
“By default, we store conversation activity for 18 months.”
Source: Microsoft Support: Privacy FAQ for Microsoft Copilot
Perplexity, in Perplexity's words
“Yes, we use this data to improve the search experience for our users.”
Source: Perplexity Help Center: What data does Perplexity collect about me?
“If you wish to opt out, go to your profile, navigate to the AI Data Usage toggle under Settings, and turn it off to stop Perplexity from using your search data to enhance our AI models.”
Source: Perplexity Help Center: What data does Perplexity collect about me?
Grok, in xAI's words
“We may use your content and interactions with Grok (e.g., prompts, searches, and other materials you submit) along with Grok's responses to train our models.”
Source: SpaceXAI Consumer FAQs (Grok)
“On the mobile app, to select or deselect using your content for model training, go to Settings, Data Controls, and select or deselect “Improve the model.””
Source: SpaceXAI Consumer FAQs (Grok)
“When using Private Chat, where available, your content and interactions are not used for model training.”
Source: SpaceXAI Consumer FAQs (Grok)
“If you delete conversations from your account or if you use Private Chat, conversations will be removed from our systems within 30 days, unless they have been de-identified or pseudoanonymized and disassociated from your account or we have to retain them for safety, security, or legal reasons.”
Source: SpaceXAI Consumer FAQs (Grok)
Meta AI, in Meta's words
“When you chat with AI, Meta may use the messages to it to train the AI model, helping make the AIs better.”
Source: Meta Privacy Center: information from chats with AIs
DeepSeek, in DeepSeek's words
“To improve and develop the Services and to train and improve our technology, such as our machine learning models and algorithms.”
Source: DeepSeek Privacy Policy
“the right to opt-out of using your Personal Data for training our models or optimizing our technologies.”
Source: DeepSeek Privacy Policy
“We retain Personal Data for as long as necessary to provide our Services and for the other purposes set out in this Privacy Policy.”
Source: DeepSeek Privacy Policy
“To provide you with our services, we directly collect, process and store your Personal Data in People's Republic of China.”
Source: DeepSeek Privacy Policy
Le Chat (now Vibe), in Vibe's words
“Allow your interactions to be used to train our models”
“Vibe: users are not opted out by default and can opt out at any time in their settings.”
Poe, in Quora's words
“In most cases, no. Only third-party developer bots and apps are permitted to use your chats to train their models.”
Source: Poe Privacy Center
“You can check the bot’s or app’s privacy shield icon to find out what type of bot or app it is.”
Source: Poe Privacy Center
“Each provider sets their own retention schedule (e.g. OpenAI and Anthropic retain your chats for 30 days).”
Source: Poe Privacy Center
More AI data
Questions
How was this scored?
It's each company's own words, quoted from its privacy or help pages on the day we checked, with a link. Our one-line summary only restates the quote. Where a page says nothing (a retention period, an opt-out), the row says so.
Does llmwise train on my chats?
No. llmwise's Privacy Policy says it doesn't train AI models on your content, and it asks the model routers it uses not to send your requests to providers that train on or keep your data for their own use. Chats stay until you delete them.
Claude, GPT, Gemini, DeepSeek, Grok, Kimi, and GLM, in one chat.
See what a message costs before you send it. Free is 5 messages to try; sign in with an email link, no password or card.