Granola
Granola
Careful
Granola says third parties such as OpenAI and Anthropic are not allowed to train on your data. Granola itself may use aggregated, de-identified data for its own models when you are opted in, and the privacy policy also says you can opt out in account settings. Enterprise training is off by default and can be turned on. The individual default is not stated cleanly, because the page talks about both opt-in and opt-out. Check the setting, and prefer the Enterprise default for a company rollout.
Last checked 2026-11-10
Data-safety checklist
Trains on your data by default
Unknown
The policy says third parties are not allowed to use personal data for their model training, and that Granola uses de-identified, aggregated data when you are opted in. It also says you can opt out in settings. Enterprise users have model training off by default and may opt in.How to opt out
Steps
Use the model-training control in Granola account settings. Enterprise admins start from off and would have to opt in.Data retention
What they keep
De-identified data that has already been built into a model is not removed, because Granola says that is not technically possible. A retention period for raw meeting audio is not stated in the section we read.BAA / HIPAA
Unknown
The privacy policy text we read does not offer a business associate agreement.Certifications
1 listed
- Unknown
Enterprise and admin controls
Controls
Enterprise model training is off unless the organization opts in. Individual accounts use the in-product setting.Sources
1 pages