Major technology companies offer privacy settings that can limit whether conversations, activity or other user content is used to improve artificial intelligence systems. The controls differ by company and product, and turning off AI training does not necessarily delete information that has already been collected.
OpenAI
ChatGPT users can open Settings, select Data Controls and turn off Improve the model for everyone. OpenAI says that new conversations will then not be used to train its models, although they can remain in the user’s chat history.
Temporary Chat provides another option. OpenAI says temporary chats do not appear in chat history, do not create memories and are not used to improve its models. Temporary chats may still be retained for up to 30 days for safety purposes.
Users should also be aware that submitting feedback on a ChatGPT response can result in the associated conversation being used to improve models, even when the general training setting has been disabled.
Google Gemini
Google provides privacy controls through Gemini Apps Activity and related Google Account settings. Users can turn off Keep Activity and manage or delete stored Gemini activity.
Google says that when Keep Activity is disabled, future Gemini chats are not used to improve its AI models unless the user submits feedback. Temporary chats are also not used to train Google’s AI models and may be retained for a limited period for safety and service purposes.
The available controls can vary by account. People using Gemini through work or school accounts may have settings controlled by their organization.
Microsoft Copilot
Microsoft provides Copilot privacy controls that allow eligible users to manage whether conversation activity is used for model training. Users can access Copilot’s privacy settings and turn off training related to conversation activity or voice conversations.
Microsoft says opting out prevents future conversation activity from being used to train its AI models. However, the company notes that conversations may still be processed for other purposes, including security, safety, compliance and product-related functions.
Meta AI
Meta’s AI privacy controls vary depending on the product and how Meta AI is being used. Settings can differ between social-media services, AI assistants and AI-enabled hardware.
Meta has also introduced controls for certain AI-enabled products that allow users to manage how data collected through those products can be used. Users should check the privacy and AI settings for the specific Meta service or device they use.
Review more than the training setting
Turning off AI training does not necessarily stop a company from retaining information for security, legal, safety or other operational purposes. It also does not automatically remove information that was collected before the setting was changed.
Users seeking greater control should review AI training settings, activity history, retention controls, connected applications and feedback settings separately. Privacy options can also change as companies update their products and policies.
For sensitive information, users should avoid entering passwords, financial details, confidential business documents or other private data into consumer AI services unless they have reviewed the service’s data practices and account controls.