Table of Contents
Teams Profanity Filter Looks for Obscene or Profane Words
When I read message center notification MC537415 (5 Apr 2023, Microsoft 365 roadmap item 113412) about the “newly introduced toggle to turn on/off profanity filtering in live captions” in Teams settings (Figure 1), I wondered about how much usage this feature will get. I can imagine that it will be popular and useful in education settings (hopefully, those running the meetings will take control if profanities abound), but the corporate world is slightly more robust.
Deployment of the profanity toggle started to targeted release tenants in early April. After completing this phase, standard release tenants will see the feature in early May. Full deployment to commercial and GCC tenants should be complete by mid-May. GCC High and DoD tenants should see it in mid-July.
Filter Inserts Masks for Bad Words
The profanity toggle controls whether Teams inserts ***** masks when the transcription of a meeting to generate live captions detects that a spoken word is profane (Figure 2). In this respect, I believe that Teams looks for words deemed to be vulgar or obscene rather than irrelevant or unsanctified.
Some corporate executives I have direct experience of, including some at Microsoft, were prone to dropping a few f-bombs during meetings to emphasize a point or stress just how they felt about a discussion. This is the kind of behavior that I think Teams will seek to filter out.
If you don’t turn the toggle on, you see every word Teams generates for a caption. This is the default, so you need to turn the toggle on if you think that you’re in danger of being offended by seeing a profane caption. Of course, if you’re listening to the meeting, you’ll hear the objectionable term in its full glory because Teams doesn’t filter the audio feed.
Factors Affecting Detection
Another thing to consider is that the generation of live captions depends on the transcription engine being able to recognize words. When seeking to identify profanities, the engine must first recognize the word and then determine it to be profane. Quite how this process will deal with local accents, argots, and slang remains to be seen. My experience when testing the Teams profanity filter was that it detected some words and ignored others. Your mileage might vary depending on the clarity of enunciation by meeting participants and quality of microphones. A whispered expletive is unlikely to be picked up by live captions.
A further challenge exists for live translated captions (one of the four features moved to Teams premium in January 2023). I do not know how effective the profanity filter will be in a meeting where participant contributions are in multiple languages or if the filter will be able to handle concurrent translation of profanities from multiple languages. That would be an interesting test to run.
Improvements for Live Captions
MC538737 (April 8) notes that users will be able to choose colors, height, position (left or bottom) and font size for live captions. There’s no facility to highlight profanities in a different color. This update will roll out in mid-May 2023.
Opting to use the Teams profanity filter is very much a personal choice. Some won’t care at all if someone’s contribution is on the edge while others will be very offended at terms that some consider unremarkable. There’s no way to set the control programmatically with PowerShell or the Graph that I can see to allow administrators to enable the filter for users. Maybe it’s best to leave this option for individuals to make up their own mind.
Support the work of the Office 365 for IT Pros team by subscribing to the Office 365 for IT Pros eBook. Your support pays for the time we need to track, analyze, and document the changing world of Microsoft 365 and Office 365.