Otter.ai
Otter says it can train proprietary AI on de-identified recordings and transcripts, while an account feedback-and-training setting controls whether conversations are shared for training, product improvement, and human review.
De-identified audio and transcripts may be used to train proprietary models. The Help Center says 'Allow sharing' gives Otter and providers access to account conversations for training and improvement. Imported Google Workspace documents are stated not to be used for training.
Open Account Settings -> Meetings -> Feedback and Training and set the permission to something other than 'Allow sharing.' The setting can be changed at any time.
Otter says it uses a proprietary de-identification method before training. It also says third-party AI providers do not store data sent through their APIs or train their own models on customer data.
Deleted conversations remain in Trash for 30 days unless permanently deleted earlier. Organisation administrators can request custom retention periods of at least 24 hours.
Human review of conversations is possible when the account grants 'Allow sharing.' Otter says the automatic de-identified training method itself does not manually review recordings or transcripts.
Foundational reference pages for teams adopting, deploying, and operating production AI with inspectable controls.