How do I ensure my AI transcription tool isn’t training on my client’s data?

You must explicitly opt-out in your account privacy settings, or upgrade to an enterprise plan that features strict zero-data-retention policies.

How do I ensure my AI transcription tool isn’t training on my client’s data? You must actively navigate the privacy settings of your software, as almost all consumer-tier AI products opt you into data training by default. If your tool trains on client data, you risk breaching confidentiality agreements and exposing proprietary information.

Here’s the framework to make sure your client data remains untouched:

Check the Terms of Service (ToS)
Do not assume privacy. Search the provider’s ToS for “machine learning,” “model training,” or “data usage.” If they reserve the right to anonymize and use your data, you must take action.
Toggle the Opt-Out Setting
In platforms like OpenAI (ChatGPT Enterprise) or Grammarly, you must go to Settings > Data Privacy and manually uncheck the box that says “Improve our models.”
Switch to Enterprise or API Solutions
Most companies do not train on data passed through their paid APIs. For example, OpenAI’s API strictly prohibits training on customer data by default, whereas their free web interface uses your inputs.

Here’s what you should do

Log into your transcription provider’s dashboard right now. Locate the “Privacy” or “Data Control” tab. If there is no explicit toggle to disable AI training, cancel your subscription and migrate to a provider like AssemblyAI or Deepgram, which offer strict data protection and do not train on customer API payloads.