Privacy grade breakdown
Weighted from the six policy factors below. The formula lives in the methodology.
Policy details
What the company’s own documents state, with the nuance kept.
Default model training
Consumer Gemini Apps train on future chat content, including text and attachments, unless the training setting in Gemini History is turned off. History retention is controlled separately (auto-delete 18 months by default). Human reviewers may review chats to improve Gemini and for safety; signed-out use is covered by the Google Privacy Policy.
Opt-out mechanism
In Gemini History (myactivity.google.com/product/gemini), turn off the setting that controls whether chat content, including text and attachments, is used to train generative AI models. Retention and training are separate settings, so history can be kept while training is off. Google is rolling this out globally; accounts still showing only the legacy 'Keep Activity' setting must turn that off to stop both retention and training.
Human review of chats
A subset of chats is reviewed by human annotators to improve models, reduce unintended activations, and keep Gemini safe. Reviewers assess response quality, accuracy, and safety. Turning off Keep Activity stops future chats from being reviewed for improvement.
Third-party sharing
Shared with service providers for human review (disconnected from account first). Connected third-party apps receive data per their own privacy policies. Google does not sell personal data.
Data retention and deletion
Gemini History auto-deletes after 18 months by default, adjustable to 3 months, 36 months, or indefinite. Temporary chats and History-off chats are retained around two months. Reviewed chats and related data are kept up to 3 years.
Delete individual chats or clear all Keep Activity anytime. Chats already reviewed by human annotators are retained up to 3 years disconnected from your account.