TikTok announced voice comments — audio messages up to 60 seconds, attached directly under any video — on September 3, 2026, rolling out globally over the following month to accounts 18 and older. The problem: every comment-monitoring tool brands rely on for branded content, TikTok's own keyword filter included, is built to scan text. A voice comment simply doesn't get read. If a viewer says out loud, under a sponsored post, something a brand would want flagged — a competitor's name, an unverified safety claim, a question about missing disclosure — none of today's filters catch it. If you're running branded content on TikTok, your comment-monitoring workflow needs a line for voice comments before your next post goes up.
What did TikTok just change about comments?
TikTok's newsroom post "Say More in the Comments," published September 3, 2026, introduced four new comment formats at once. Voice Comments let anyone tap a microphone icon in the comment box and record up to 60 seconds of audio; anyone reading the thread taps to listen instead of read. Comment Polls let creators put up to five options directly in a comment and track results without leaving the app. Photo Carousel Comments allow up to nine images in a single comment, and Live Photo Comments attach a brief burst of motion to a still image. Of the four, Voice Comments is the one that matters for brand safety, because it's the only format that moves commentary out of a medium every monitoring tool — TikTok's own and every third-party vendor's — was built to read.
Why does this create a brand-safety blind spot?
Every comment-safety tool available on TikTok today operates on text. The platform's own keyword filter, tucked under Settings → Privacy → Comments, works from a custom word list, and practitioner guides to the feature consistently note two limits: the list tops out well short of what a brand managing multiple products, competitors, or languages typically needs, and matching is exact — no wildcards, no regex, no catching a deliberate misspelling. Layered on top of that, TikTok's "Filter spam and offensive comments" toggle and its newer Smart Keyword Filters use AI to catch synonyms and context, but that AI is still classifying text strings. Third-party comment-monitoring vendors that brands bring in for high-volume campaigns — the kind that watch for competitor mentions, unverified claims, or disclosure complaints at scale — run the same way: they ingest the comment text and score it. None of these tools transcribe audio. A voice comment isn't missed by the filter; it was never something the filter could see in the first place.
How does this compound the stitch/duet gap that already existed?
TikTok already has a known comment-moderation gap that predates voice comments: the platform does not apply a creator's own comment filters to stitches and duets of their video, so a branded post that takes off via reposts effectively spawns comment sections a brand's monitoring setup was never configured to watch in the first place. Voice comments now layer a second gap on top of the first, inside the very comment sections brands already do monitor. On a sponsored post that goes viral through stitches and duets, comment volume balloons across sections the monitoring plan didn't account for, and within every one of those sections — original and reposted alike — a slice of the replies will now arrive as audio no keyword tool reads either way.
Example (a hypothetical to size the problem, not a published adoption figure): say a branded post draws 800 comments and 5% come in as voice notes. That's 40 comments that no keyword filter, no AI toggle, and no third-party monitoring vendor will ever flag — the only way to know what's in them is for a person to press play on each one by hand.
What should you check before your next branded post?
Before your next branded or sponsored TikTok post goes live, confirm these with whoever owns comment monitoring — the creator, an agency, or an in-house team:
- Who is listening, not just reading. Confirm the monitoring workflow includes someone periodically tapping to listen to voice comments, not only scanning text.
- Whether your vendor transcribes audio. If a third-party comment tool is part of the setup, ask directly whether it processes voice comments at all — most comment-safety products on the market today were built before this feature existed.
- Whether stitches and duets are in scope. If the post is likely to spread that way, confirm monitoring covers those comment sections too, not just the original post.
- Whether the campaign's audience is under 18. Voice comments are limited to accounts 18 and older for now, so campaigns aimed at younger audiences see less immediate exposure — but don't assume an account's stated age matches its actual audience.
What should you add to briefs and vendor contracts now?
The categories brands already treat as sensitive — disclosure questions, safety or efficacy claims, pricing and stock questions, complaints — don't stop being sensitive because they arrive as audio instead of text. Update the brief and any monitoring vendor agreement to say so explicitly:
- Name voice comments in the monitoring SLA. If the brief or contract currently says "monitor comments," specify that this includes voice comments, not just the text thread.
- Set a review cadence for audio, even a sampled one. A brand doesn't need to listen to every voice comment on a viral post, but a defined sampling routine (say, a set number of voice comments reviewed per shift) beats no routine at all.
- Route the same sensitive categories to the brand regardless of format. If your brief already routes disclosure or safety questions to the brand instead of letting a creator or their AI assistant reply — the same issue we flagged when Facebook's AI began drafting comment replies — extend that routing rule to anything raised in a voice comment too.
Once the monitoring plan actually covers what's being said — spoken or typed — the next question is whether the content built around it is converting. Hyperstar tracks live creator content and its sales contribution in one dashboard, across every format a platform ships. Get started.
