How do I mask sensitive and personally identifiable information (PII) in files?

Marvin can automatically detect and remove any mention of personally identifiable information (PII) in transcripts, audio, and video recordings. PII includes names, email addresses, and other identifiable information – such as age, gender, race, etc.

You can also mask PII in surveys and support tickets. Learn more about this in our help center article.

Use PII redaction to ensure that your research complies with data privacy regulations like GDPR and HIPAA.

Marvin can mask sensitive information for recordings and transcripts in the following 30 languages:

1. English (US)

2. English (GB)

3. English (AU)

4. English (IE)

5. English (IN)

6. English (AB)

7. English (WL)

8. French (FR)

9. German

10. Portuguese (PT)

11. Spanish (US)

12. Dutch (NL)

13. Polish (PL)

14. Spanish (ES)

15. Vietnamese

16. Italian (IT)

17. Hindi (IN)

18. Japanese (JP)

19. Chinese (CH)

20. Finnish (FI)

21. Korean (KR)

22. Russian (RU)

23. Turkish (TR)

24. Ukrainian (UA)

25. French (CA)

26. German (AT)

27. German (CH)

28. Portuguese (BR)

29. Spanish (Colombia)

30. Spanish (Mexico)

Navigate to your project from Home or Projects.

Go to the Files tab in your project and hover over the file you want to redact. Select Hide participant's personal info against the file name. You'll see this option for every file.

You'll also see this option within the file. Click All actions in the top toolbar and select Hide participant's personal info.

You'll see a range of options and controls to mask PII. Let's explore how each of them works:

Note: If any notes were added to the transcript before redaction, the notes will remain unchanged. Please check and create notes carefully for such files.

Once you’ve selected the changes you want to make, click Apply redaction.

Marvin securely deletes the original video files after processing and stores just the PII-redacted version. This action is not reversible.

Marvin will redact words such as "Claude" because it appears to be a person’s name. However, knowing that a participant prefers using Anthropic’s Claude over other AI tools may be information crucial to your study.

You can specify words and phrases you want Marvin to retain when it runs automatic redaction. You can create this list of words excluded from redaction to be applied to all files in a project, as well as at the team level, i.e., all projects in Marvin. At the moment, you cannot create exclusion lists for just one or a few files.

Only Admins can create a PII exclusion list at the team level. They must select Settings > Admin settings from the left navigation menu.

Click Data privacy > Edit settings.

Against Exclude words from redaction, select Show list to create or edit the list of words.

Your company name is added to the list by default. You may edit or delete it from the list.

To separate multiple terms, words, or phrases, you can use any of these formats: add commas or semicolons, add new lines, press Tab, or press Enter. You can also paste a list you’ve already prepared. Click Continue once done.

Remember to select Save settings to apply these changes.

Once updated, Marvin will not redact these terms or phrases in any file in any project. These terms will appear in transcripts, audio, video, open-ended survey responses, and meeting recordings. If you make changes to the list, those will not be applied to files that have already been redacted. You must run PII redaction again for such files.

Admins, Full seat users, and Collaborators with access to a project set up PII exclusion if they have access to a project. They can also add the list while creating a new project.

Open your project and select Project settings > Data Privacy from the left navigation menu. Click Edit settings/Override and edit settings to access the PII settings.

Select Show list against Exclude words from redaction to create or edit the list of words.

If a term has been excluded for redaction at the team level, you will see it here too. However, it can also be edited or removed from Admin settings.

To separate multiple terms, words, or phrases, you can use any of these formats: add commas or semicolons, add new lines, press Tab, or press Enter. You can also paste a list you’ve already prepared. Click Continue once done.

Remember to select Save settings to apply these changes.

Once updated, Marvin will not redact these terms or phrases in all files in the project. These terms will appear in transcripts, audio, video, open-ended survey responses, and meeting recordings. If you make changes to the list, those will not be applied to files that have already been redacted. You must run PII redaction again for such files.

Marvin excludes all variations of your specified words, including if they are

If "New" and "New York" both are on the list, the entire term New York will be redacted wherever there might be a doubt.

If only "John" is excluded but the text says "John Smith"), Marvin will retain Jon and redact the rest, shown as "John ██████."

PII redaction may take between a few minutes to a few hours, depending on the duration of the interview and the PII you want to mask.

Admins can choose to let file owners (the person who recorded with Marvin or uploaded the file to Marvin) access and work on the file as Marvin redacts data in the background. Admins can enable this from the "Data privacy" tab in Admin settings. They must toggle on the setting that says "All file owners to view and annotate while sensitive data is being redacted."

Since file owners already recorded or uploaded the file before PII redaction was enabled, they already know the sensitive information in the file. Hence, Admins can let them make notes and read the transcript even during compliance processing.

However, this is only available to the file owner and not everyone who has access to the project. All other users with access to the project will see this message: Redacting information.

Admins may also choose to disable this setting.

File owners have limited access till the processing is complete. Here are some of the actions they can take:

As soon as the file is processed, all actions are enabled for everyone with access to the project.

Do note that if you make a note or run Ask AI on data that will be scrubbed, the note and answer will remain, but the original content will be redacted.

During redaction, file owners will not be able to:

You can mask PII in bulk rather than repeating this process for each file. Select all or multiple files by clicking the check box on each file or in the toolbar. To select multiple files, use the keyboard shortcut Command + Click. (You can click anywhere on the file and not just on the check box).

Then select Hide participant's personal info from the three dots in the toolbar.

You can set up Marvin to automatically blur video, mask voices and redact text in transcripts for all recordings in all new projects.

Head over to Admin settings, and select your preferences under "Data privacy"

Your enabled settings will apply to all new projects. In case you want to change the settings for a particular project, you can override this from individual project settings. You can set this up for a new project and an existing one.

That's it! You can now share and store your research without revealing any sensitive information.

Q. Does Marvin also mask PII when someone shares their screen during a call?
A. Yes, you can blur a specific part of the video. Marvin can also detect the PII visible on a screen share.
​
​Q. Can I redact PII for AI moderated interviews?

A. Yes, all PII features are available for interviews run by the AI moderator.