"Does this chatbot remember what I type, and does it train on it?" The short answer: it depends on which account you use and which settings are on. The longer answer decides whether your work data ends up in a future AI model. Here is how it works, and what you can do about it.
First, the key distinction: account, not tool
Whether an AI tool uses your conversations for training depends less on the tool and more on the account type:
- Business accounts. ChatGPT Business, Enterprise, and Edu do not use workspace data to train models by default. Microsoft 365 Copilot falls under commercial data protection and does not use your business data for training.
- Personal and free accounts. Here, conversations are more often used to improve the model, unless you turn that off yourself.
In other words, the same tool can carry different risks in a free account than in a managed business environment.
How to turn off training per tool
Settings change regularly, so always check the current menu. The place to look is always the privacy or data settings.
- ChatGPT. Go to Settings, then Data Controls, and turn off the option to improve the model for everyone. For one-off conversations you can use Temporary Chat: it does not appear in your history, does not create memory, and is not used for training. OpenAI does note that copies may be kept for up to 30 days for safety purposes.
- Google Gemini. Manage your Gemini Apps Activity in your Google account and turn it off if you do not want conversations kept and potentially reviewed by humans.
- Anthropic Claude. Check the privacy settings to see whether your consumer conversations may be used for training, and set it to your preference.
- Microsoft Copilot. In the business 365 context, your company data is not used for training; in the consumer version, check the privacy settings.
For details per tool, see the individual guides, for example on ChatGPT, Gemini, or Copilot.
Deleting a chat: what it does and does not do
Deleting a conversation removes it from your history and prevents that specific text from being used for further training. What it does not do: retrieve data already used to train an existing model. Once "baked into" a trained model, information is no longer in a box you can empty. So turning off training and deleting chats mostly help going forward, for new conversations.
The right to erasure under the GDPR
As a European user you have the right to erasure (Article 17 GDPR), also known as the right to be forgotten. You can submit a request through the provider's privacy portal, for example to delete your account and stored conversations. Keep in mind that this primarily covers stored data and future processing, and that providers may retain certain data where there is a legal basis to do so.
Why opt-out does not solve the problem
Turning off training is sensible, but it does not solve the core risk for work. Two reasons:
- Opt-out is about what the provider may do, not about what you share. Even without training, a sensitive detail can be stored, readable by a connected app, or travel further than intended through a shared link.
- A personal account with opt-out is not an approved work environment. For customer, patient, or case data, a managed business environment with a data processing agreement is the starting point, not a toggle in a free account.
So the most important control is not the setting. It is what you put in the prompt. See what you can and cannot share with AI.
The data you do not share needs no opt-out at all
Opt-out limits what happens to your data after it is sent. It is more effective to keep sensitive data from going in at all. BeeSensible highlights personal data while you type in browser-based AI tools, so you can remove, replace, or mask it before you send. What does not leave the prompt does not need to be turned off, deleted, or requested back.
Further reading: AI data leakage and data sovereignty and AI prompts.
