Is Claude private? It depends on the plan and one setting. On Claude Free, Pro and Max, Anthropic can use your chats to train its models if "Help improve Claude" is switched on, and then keeps them for up to five years. With the setting off, chats are kept for 30 days. Claude Team, Claude Enterprise and the API run under Anthropic's commercial terms: not used for training by default, with a data processing addendum included.
Below: what changed in 2025, how to check your setting, what Anthropic keeps in each case, and what that means for confidential work.
What changed in 2025
In August 2025, Anthropic updated its consumer terms and privacy policy. Before that, consumer chats were not used for training by default. Since the update:
- Claude Free, Pro and Max users choose whether their chats may be used to improve Anthropic's models.
- Existing users saw a pop-up and had to make a choice by 8 October 2025 to keep using Claude. The training toggle in that pop-up started switched on, next to a large "Accept" button.
- If you allow training, retention for new and resumed chats extends from 30 days to up to five years.
- Business products (Claude Team, Claude Enterprise, Claude Gov) and API access, including through Amazon Bedrock and Google Cloud Vertex AI, were not affected.
If you clicked through that pop-up quickly, it is worth checking what you agreed to.
How to check and change the training setting
- Click your name in the bottom left corner of claude.ai or the desktop app.
- Open Settings, then Privacy.
- Find Help improve Claude and switch it off.
The change applies to new and resumed chats and coding sessions, including Claude Code on a personal Pro or Max plan. Chats you deleted are not used for future training.
Incognito chats (the ghost icon) are never used for training, even if the setting is on, and are not saved to your chat history or to Claude's memory.
What Anthropic keeps, by plan
| Plan | Used for training | How long chats are kept | Data processing agreement |
|---|---|---|---|
| Free, Pro, Max, training on | Yes | Up to 5 years (de-identified) | No |
| Free, Pro, Max, training off | No | 30 days | No |
| Incognito chat (consumer) | No | Not kept in history | No |
| Claude Team | No, by default | In history until you delete it; deleted data removed from the back end within 30 days | Yes, under the commercial terms |
| Claude Enterprise | No, by default | Indefinite by default; admins can set a custom period (minimum 30 days) | Yes, under the commercial terms |
| API | No, by default | Short back-end retention; zero data retention available by agreement | Yes, under the commercial terms |
Based on Anthropic's privacy centre and consumer and commercial terms as of November 2026. Check the current version for your plan.
Three exceptions apply on consumer plans regardless of the toggle:
- Feedback. If you rate a response or report it, Anthropic can keep the related conversation for up to five years and use it for improvement.
- Safety flags. Conversations flagged for a usage policy violation are kept longer under trust and safety rules.
- Deletion takes time. A deleted chat leaves your history at once and Anthropic's back-end systems within 30 days.
Claude Enterprise retention controls
On Enterprise, the Primary Owner or an Owner sets retention under Organization settings > Data and Privacy. Without a custom period, conversations, projects and artifacts are kept indefinitely so users can see their history. Projects keep their own retention, so chats inside a project are not removed by the chat retention period. Zero data retention for Enterprise is arranged with Anthropic's account team rather than switched on by default.
Claude privacy and GDPR
For anyone in the EU handling personal data, the plan matters more than the toggle:
- Consumer plans (Free, Pro, Max) come without a data processing agreement. Anthropic acts as the controller of that data, so using them for client or patient information is hard to justify under Art. 28 GDPR.
- Team, Enterprise and the API include Anthropic's data processing addendum through the commercial terms, which is the starting point for business use.
- Where data is processed is a separate question. Through the API on Amazon Bedrock or Google Cloud Vertex AI, Claude can be called in EU regions. Check what applies to the product you use.
For professions with a duty of confidentiality (lawyers, doctors, therapists, tax advisers), a data processing agreement and no training are the minimum, and short or zero retention helps reduce the risk further. Our GDPR AI compliance checklist lists the other points to check.
How Claude compares with ChatGPT and Gemini
All three consumer services let you opt out of training, and all three keep something afterwards: ChatGPT Temporary Chats for up to 30 days, Gemini chats for up to 72 hours with Keep Activity off, Claude chats for 30 days with training off. The step-by-step for each is in how to stop ChatGPT, Gemini and Claude training on your data, and the wider comparison is in ChatGPT vs Claude vs Gemini privacy.
Using Claude without the trade-offs
Wysor gives you Claude alongside GPT, Gemini, Grok, Mistral, Perplexity and open-source models in one EU-hosted workspace.
- Zero data retention with the model providers, so your prompts are not stored by Anthropic or the others.
- No training on your data, on any plan, written into the contract.
- Hosted in the EU on AWS in Frankfurt, with a data processing agreement included.
- Same Claude models, and you can switch to another model in the same chat.
There is a free plan, Plus at €19.99 per month and Premium at €29.99 per user per month. If you are weighing options, see our Claude alternatives overview.
Try Claude in Wysor for free →
Frequently asked questions
Is Claude private? Partly, depending on plan and settings. On consumer plans with "Help improve Claude" off, chats are not used for training and are kept for 30 days. With it on, they can be kept for up to five years. Team, Enterprise and the API are not used for training by default and include a data processing addendum.
Does Claude train on my conversations? On Claude Free, Pro and Max, only if "Help improve Claude" is on. Since the 2025 terms update, many users have it on because the toggle started switched on in the pop-up. Check Settings > Privacy. Incognito chats are never used for training.
How long does Claude keep my chats? 30 days with training off, up to five years with training on, and up to five years for conversations you sent feedback on. Deleted chats leave Anthropic's back-end systems within 30 days.
Is Claude Pro more private than Claude Free? No. Free, Pro and Max follow the same consumer terms and the same training setting. Paying for Pro does not change how your data is handled.
Is Claude Team private enough for client data? Claude Team is not used for training by default and includes a data processing addendum, which covers the basics. Chats stay in history until deleted, so retention is in your hands. Whether that is enough depends on your profession's rules and the data involved.
Is Claude GDPR compliant? No AI tool is compliant on its own; it depends on how you use it. Consumer plans offer no data processing agreement, so they are a poor fit for personal data. Team, Enterprise and the API include one, which is the basis for GDPR-conscious business use.
