Most people leave Otter.ai for one of two reasons. The ordinary one is limits: the free tier gives you 300 minutes a month and caps a single conversation at 30 minutes, which a weekly team meeting eats through by the third week. The other reason is the one that has been getting louder — a growing discomfort with a bot that joins your calls, records them, and keeps the recording on someone else’s servers.

That discomfort now has a court case attached to it. In August 2026 a federal judge allowed wiretap and biometric-privacy claims against Otter.ai to proceed, on the theory that the service is a third-party eavesdropper rather than a participant. Nothing has been proven, and that distinction matters — but it is worth understanding before you decide where your meetings should be processed. This guide covers what the ruling actually said, what Otter does with your audio, and how the same job works when your audio never leaves your machine.

Disclosure: I make Samurai Scribe, one of the apps below. I have tried to describe the others the way I would want mine described, and every claim about them links to their own pages or to primary sources, checked in October 2026.

What the Otter.ai ruling actually said

On 13 August 2026, Judge Eumi K. Lee of the U.S. District Court for the Northern District of California ruled on Otter’s motion to dismiss in In re Otter.AI Privacy Litigation (No. 5:25-cv-06911), granting it in part and denying it in part.

What survived and moves toward discovery:

the federal Wiretap Act claim,
the California Invasion of Privacy Act (CIPA) claim,
the Illinois biometric-privacy (BIPA) claims, concerning voiceprints.

The reasoning that matters for the whole category: the court found plaintiffs had plausibly alleged Otter acts as a third-party eavesdropper rather than an invited participant, because it retains recordings and uses them for its own purposes, including training its speech-recognition models. Some claims were dismissed, including the CFAA and Washington Privacy Act claims, and several common-law privacy claims where plaintiffs had not pleaded specific facts showing a reasonable expectation of privacy.

Two honest caveats. First, this is a pleadings ruling: the court accepts the complaint’s allegations as true at this stage and has made no finding that Otter violated any law. Second, and more practically — this changes nothing about your obligations. A local app removes the third party from your call. It does not remove your duty to tell the people in the room that you are recording. That is not legal advice, and the rules genuinely differ by state and country.

Sources: the order as covered by the National Law Review and Sheppard Mullin.

What Otter does with your meetings

Otter is a cloud service, which is the source of both its convenience and the question above. A bot joins your Zoom, Teams or Meet call — or, with Otter’s newer desktop app, your computer records without one — and either way the audio is transcribed on Otter’s servers and the transcript lives in your Otter account.

From their pricing page, as of October 2026:

Plan	Price	Transcription	Per conversation
Basic	Free	300 min/user/month	30 minutes
Pro	$16.99/user/mo, or $8.33 billed annually	1,200 min/user/month	90 minutes
Business	$30/user/mo, or $19.99 billed annually	Unlimited meetings	4 hours
Enterprise	Not listed	Unlimited	4 hours

Minutes are a shared pool and do not roll over. The free tier also allows three file imports, ever.

None of this is a scandal; it is how a cloud product works, and it is the same shape as Fireflies, Fathom and most of the category. But it does mean the answer to “where is my meeting audio right now?” is “on a server I don’t control”, and for some work — client calls, therapy sessions, interviews under embargo, anything under an NDA — that answer is the end of the conversation.

The difference is easier to see than to describe:

Both produce a transcript. They differ in whether anything crosses the edge of your machine.

What “offline” should mean before you trust it

“Private”, “secure” and “on-device” get used loosely. Four things are worth checking in any tool that claims them:

It works with the network off. Disconnect, record a meeting, and see whether you get a transcript. This is a one-minute test and it settles the question.
No bot joins the call. Audio captured at the operating-system layer means nothing appears in the participant list and nothing needs a seat in your meeting.
The models are on your disk. Local transcription means a model you downloaded once, not an API call.
The output is yours. Files you can read without the app, in a folder you chose, so leaving costs you nothing.

Local alternatives to Otter, for Windows and Mac

Checked October 2026, from each product’s own pages.

App	Price	Platforms	Where audio goes	Speaker labels	Notes stored as
Samurai Scribe	Free · Pro $19 once (beta; $39 list)	Windows, Mac (Apple Silicon)	Your computer	Yes, free tier, including in-room	Markdown in a folder you choose
Meetily	Free (open source) · Pro $10/user/mo, billed $120/yr	Windows, macOS, Linux from source	Your computer	Diarization is Pro	Local; Markdown export is Pro
MacWhisper	Free · Pro €64 once	Mac only	Your Mac	Yes	Transcripts in app; exports
Granola	Free (30-day history) · Business $14/user/mo	Mac, Windows, mobile	Cloud (Deepgram/AssemblyAI, then OpenAI/Anthropic)	Not stated	Granola’s cloud
Otter.ai	Free 300 min/mo · from $8.33/user/mo	Web, Mac, Windows, mobile	Otter’s cloud; a bot, or its desktop app	Yes	Otter’s cloud

If you are on a Mac and mostly transcribe files you already have, MacWhisper is excellent and I would not try to talk you out of it. If open source is the deciding factor, Meetily is the honest answer — ours is not open source. What we do differently is Windows and Mac both, speaker labels on the free tier including people sharing one room, and notes written as plain Markdown rather than exported from a database.

Moving off Otter, step by step

1. Take your transcripts with you first

Export from Otter before you cancel; a lapsed account is an awkward place to discover you needed something. Plain text or Markdown exports drop straight into a vault folder.

Old recordings you have as files — or a webinar you only have a link to — import and transcribe the same way; there is a walkthrough in transcribing YouTube videos and lectures offline.

2. Install and pick a model

Download it for Windows 10/11 64-bit or an Apple Silicon Mac — no account, no email. The Windows installer is code-signed, so Windows shows a verified publisher; on a brand-new release SmartScreen may still ask you to confirm. Setup downloads one speech model, once.

3. Record the meeting without a bot

Audio is captured from your microphone and system output at the OS layer, as two separate channels with echo cancellation, so a Zoom or Teams call is recorded without anything joining it — and an in-person conversation works the same way, which is the case Otter’s bot cannot cover at all. How to set this up for Zoom, Teams and Meet, and what can go wrong with recording this way, is in an AI notetaker without a bot.

4. Read who said what

A local diarization pipeline separates speakers, including several people sharing one microphone in a room. You can rename speakers once and have it apply across the transcript.

5. Keep the output as files

Every meeting becomes a folder: notes, transcript, summary and the recording. Obsidian vaults are detected automatically. Newer versions also write an AGENTS.md into the vault, so tools like Claude Code can read and propose edits to your notes with your review — which is only possible because the notes are files in the first place.

For the Windows-specific walkthrough, see a Granola alternative for Windows that works offline. On a PC coming from a Mac-only tool, a MacWhisper alternative for Windows covers the same ground. If dictation rather than meetings is what you need, there is a Wispr Flow alternative that works offline.

Removing the third party does not remove your responsibility. Consent rules depend on where you and the other participants are — some places require everyone’s agreement, some only one party’s. Say you are recording. A local tool means no company other than you holds the audio; it does not make a recording lawful on its own, and this is not legal advice.

FAQ

Does it really work with no internet? Yes, after the first run downloads a model. Transcription, speaker labels and summaries all run on your CPU or GPU. Left alone the app still makes a daily update check and, if you bought Pro, a licence check — both listed on the network disclosure.

Can I import my old Otter recordings? Yes. Common audio and video formats import and transcribe like any other meeting.

Does it work on Windows? Windows 10 and 11, 64-bit, and Apple Silicon Macs. No Intel Mac build yet.

What about in-person meetings, not calls? That is the case this handles best: one microphone in a room, with speakers separated on the transcript. Otter’s meeting bot can’t cover it; its mobile and desktop apps can record a room, but the audio still goes to Otter’s cloud.

Is it free? The free tier has no time limit and no minute cap, and includes speaker labels, summaries and the assistant, with the smaller Whisper models and NVIDIA Parakeet. To hear a Pro transcription model on your own audio first, download it in Settings → AI Models → Transcription and try it in Compare models. Pro is $19 once during the beta ($39 after) for the larger Whisper models, Parakeet Ultra, live meeting help and skills that run by themselves.

Is it open source? No. Meetily and a few others are, and if that is your deciding factor you should use one of them. What we offer instead is verifiability: the network disclosure, and output you can read without us.

Do I have to leave Otter entirely? No, and plenty of people won’t. If what you need is a searchable archive your whole company shares, a cloud tool is built for that and this isn’t. The case for running it locally is narrower and simpler: some conversations should not be on anyone else’s computer, and those are the ones worth moving first.