Opting out of model training

Turn off the switch that lets Didit use your organization's verification data to improve its models. It lives in Settings on the Account tab, under Privacy and data use. One organization-wide control, effective immediately, and not the same setting as data retention.

Written by Didit Support · Updated 9 September 2026

Didit allows model training by default. Any organization owner or administrator can opt the whole organization out at any time, without contacting anyone.

#Opt out from the console

1. Open the Account tab

   Go to Settings in the Business Console and stay on the Account tab, the first one. It is the organization page, the one that also holds your organization name and the list of applications.

2. Find the Privacy and data use section

   Scroll to the Privacy and data use section, just below your organization details.

3. Turn the switch off

   Turn off Allow Didit to use my organization's verification data to improve models.

4. Confirm

   A dialog asks Opt out of model improvement?. Click Opt out. The section then shows an Opted out badge.

#It is one switch for the whole organization

There is no per-application version of this setting, and it does not live under an application's settings. One switch on the Account tab covers every application in the organization, and every verification your organization has already run.

#What opting out does

Once the change takes effect, Didit excludes both your organization's historical and newly collected data from future:

• model training and fine-tuning;
• model evaluation and validation;
• dataset curation; and
• training-data exports.

Each effective change is recorded for audit purposes.

#What opting out does not do

• It does not delete anything. Your verification records stay exactly as they were. To remove data, see deleting sessions and personal data.
• It does not change your retention window.
• It does not affect live processing. Verification, fraud detection, support, security and compliance processing continue unchanged.
• It is not retroactive on models. Models already trained before you opt out are not retrained or untrained.

#Who can change it

Only members with permission to update the organization, which in practice means owners and administrators. Everyone else sees the current state but cannot change it. See inviting team members and setting roles.

#Turning it back on

Turn the switch on again and your organization's eligible data becomes available for future model work. There is no confirmation dialog for opting back in.