Otter.ai’s Harder Problem Isn’t Accuracy, It’s Consent
Otter has spent a decade being the default answer to ‘what should record this call?’, and on raw transcription it remains a reasonable one. What changed between 2025 and 2026 is not the model quality but the legal weather around it. Four separate complaints filed in the Northern District of California were consolidated into a single privacy action, and every one of them turns on the same behaviour: a bot that joins meetings from a synced calendar and starts listening before anyone in the room has agreed to it.
Otter.ai plan facts, USD, checked August 2026
- Basic is free with 300 transcription minutes per user each month
- Pro is $16.99 monthly or $8.33 per user monthly billed yearly
- Business is $30 monthly or $19.99 per user monthly billed yearly
- Pro caps single meetings at 90 minutes; Business raises the ceiling to four hours
- Prices are set in US dollars and the vendor revises them without notice
What 300 free minutes actually covers
The Basic tier’s allowance sounds generous until it meets a normal working week. Three hundred minutes is five hours, which a single day of back-to-back calls will consume outright. Otter’s own plan comparison sets that allowance per user per month, and it does not roll over, so a quiet fortnight followed by a busy one leaves you short at exactly the moment the transcripts matter.
There is a second constraint that catches people later. The free tier’s per-meeting ceiling is short enough that long workshops and all-hands sessions get truncated mid-sentence, and the resulting file is not obviously incomplete on first inspection. Anyone relying on Otter as a system of record rather than a convenience should treat Basic as an evaluation tier only.
Pro at $8.33 per user per month on annual billing lifts the allowance to 1,200 minutes and caps individual meetings at 90 minutes. That ninety-minute ceiling is the detail worth checking against your calendar, because quarterly reviews, training sessions and board meetings routinely run longer, and those are precisely the conversations you most want a record of. Business removes both constraints, advertising unlimited meetings and recordings with a four-hour maximum per session.
Four complaints, one consolidated case
Reporting from the National Law Review sets out the sequence: Brewer was filed on 15 August 2025, Walker on 26 August, Theus on 3 September and Winston on 10 September, and the actions were consolidated before Judge Eumi K. Lee as In re Otter.AI Privacy Litigation, lead case number 5:25-cv-06911. The allegations are not identical but they rhyme.
- Automatic calendar-driven joining of meetings without participant consent
- Recording and transcribing people who never held an Otter account
- Capture and storage of biometric voiceprints, pleaded under Illinois BIPA in the Walker complaint
- Use of captured conversation content to train machine-learning models without disclosure
Why the default settings are the substance of the claim
Employment firm Fisher Phillips, writing for its client base rather than for consumers, framed the litigation as a warning to the businesses deploying notetakers rather than only to the vendors building them. Their reading is that the exposure sits with whoever configured the tool to join meetings automatically, because that is the party in a position to obtain consent.
That matters practically. If your organisation runs Otter with calendar sync enabled across a sales team, the recordings include customers, candidates and suppliers who never accepted Otter’s terms. In two-party consent jurisdictions that is a live risk regardless of how the case resolves, and the sensible mitigation, disabling auto-join and announcing recording verbally, costs nothing.
Where the transcript still falls over
Comparison coverage on G2 and elsewhere consistently places Otter in the same broad accuracy band as Fireflies and similar tools on clean audio from a single speaker. The divergence appears on the hard cases: heavy crosstalk, non-native accents, technical vocabulary and poor microphones.
Speaker attribution is the weaker half of the product. Otter assigns labels by voice signature, and in a six-person call with two similar voices it will merge them, which quietly corrupts any action-item extraction built on top. The summaries inherit the error without flagging it, so a wrong owner on a task reads exactly like a correct one.
The summaries compound the problem rather than containing it. Otter presents extracted decisions and action items in clean prose with no confidence indicator and no marker distinguishing a phrase transcribed accurately from one the model reconstructed. A reader skimming the recap has no way to tell which is which without opening the audio, which defeats the purpose of the recap. Treating the summary as a navigation aid pointing back at timestamps, rather than as a document to circulate unread, is the only safe working practice.
Getting your recordings back out
- Decide the export format before you accumulate a year of files, because bulk export options are thinner than the single-file ones
- Check whether your plan permits audio export as well as text, since the transcript alone loses the evidence of what was actually said
- Confirm the retention setting on the workspace rather than assuming the default matches your policy
- Disable calendar auto-join for anyone who meets external parties
- Run a deletion test on a real recording so you know how long removal actually takes
Pricing it against Fireflies and the meeting platforms
At $19.99 per user per month on annual Business billing, Otter is not cheap relative to Fireflies, whose Business tier lands at $19 annually with unlimited storage. The gap narrows further once you notice that Zoom, Teams and Meet now bundle their own summarisation into subscriptions many organisations already pay for.
Otter’s remaining advantage is that it is platform-agnostic and has a decade of interface refinement behind it. That is a real argument for a team living across three video tools. It is a much weaker argument for a team standardised on one, where the native notetaker records with consent already handled by the platform’s own disclosure.
Sources: Otter.ai pricing · National Law Review, privacy litigation targets Otter.ai · Fisher Phillips, new lawsuit highlights AI notetaker concerns · UC Today, Otter.ai on trial · Fireflies.ai pricing
