Is Otter.ai Safe? A 2026 Privacy Review

Q: Is Otter.ai safe to use?

For casual non-sensitive meetings, generally yes — an established service with enterprise security options (SOC 2, SSO). For confidential meetings, with caveats: Otter's bot joins your call and uploads audio to its cloud. A 2025 class-action lawsuit and university bans reflect concern about this for sensitive conversations. For confidential work, on-device transcription where audio stays on your Mac is safer.

Q: Does Otter.ai train AI on my conversations?

The use of user data for AI training is the central allegation in the 2025 class-action lawsuit and is being litigated. Check Otter's current data policies and opt-out settings. For confidential meetings, the surest way to avoid training-on-your-data concern is on-device transcription, where the vendor never receives your audio.

🔒🎙️

Is Otter.ai Safe?

Casual meetings: Generally OK

Confidential: Cloud + bot = risk

2025: Class-action lawsuit filed

Bans: Several universities

TL;DR: Is Otter.ai safe? For casual, non-sensitive meetings, it's a normal cloud service with enterprise security options (SOC 2, SSO on higher tiers). For confidential conversations, the picture is more complicated. Otter's bot joins your meetings and uploads audio to its cloud. In August 2025, a federal class-action lawsuit (Brewer v. Otter.ai) was filed alleging unauthorized recording and use of data for AI model training. Separately, universities including Cornell, Oxford, and Cambridge have restricted AI meeting bots over privacy. None of this means Otter is uniquely unsafe — but for legal, medical, HR, or other confidential meetings, the bot-plus-cloud architecture is exactly the exposure these institutions are reacting to. On-device transcription, where audio never leaves your Mac, avoids it structurally. Disclosure: I build an on-device alternative (MetaWhisp); this is a sourced, factual assessment.

Otter.ai safety scorecard showing normal for casual meetings but risky for confidential and regulated work with on-device no-bot as the safest option

Is Otter.ai Safe to Use?

The honest answer is "it depends what you discuss." Otter.ai is a legitimate, established company with a privacy policy and enterprise security features like SOC 2 compliance and SSO on its higher tiers. For everyday meetings that aren't sensitive — internal standups, casual calls, public webinars — it's safe in the way most cloud services are. The safety question gets sharper for confidential conversations because of how Otter works:

A bot joins your meeting. Otter adds itself as a participant in your Zoom, Meet, or Teams call to capture audio.
Audio goes to the cloud. The conversation is transcribed and stored on Otter's servers.
Everyone in the call is recorded — including people who didn't individually consent to a third party receiving the audio.

So "is Otter safe" splits: yes for casual meetings, with real caveats for confidential ones. The 2025 lawsuit and the university bans, covered below, are specifically about this confidential-meeting exposure.

"Safe" means different things and Otter passes some senses while raising others. On security — is your data encrypted and protected from outside attackers? — Otter uses standard protections and offers SOC 2 and SSO for enterprise, which is adequate. On privacy — who can access your conversations and what's done with them? — the picture is more contested, which is what the class-action lawsuit centers on. On consent — did everyone in the meeting agree to be recorded by a third party? — a bot in a group call records everyone, and not all of them necessarily consented, which is what the university bans address. Most people asking "is Otter safe" mean the first sense and Otter largely passes. But for confidential or multi-party meetings, the second and third senses matter more, and that's where the concerns concentrate. Knowing which kind of safety you actually need is the whole question.

What Is the Otter.ai Lawsuit About?

In August 2025, a federal class-action lawsuit, Brewer v. Otter.ai, was filed against the company. Per independent reporting on the case, the suit alleges that Otter engaged in unauthorized recording of conversations and used user data to train its AI models without adequate consent. A few things to understand about what this means:

It's an allegation, not a finding. A lawsuit being filed is not proof of wrongdoing; it's a claim that will be litigated. Otter can and likely will contest it.
The core concern is consent and AI training. The specific worry is whether people in recorded meetings consented to Otter capturing their voices and whether that data trained Otter's models.
It reflects a broader question about AI meeting tools generally — who consents when a bot records a multi-person call, and what happens to that data afterward.

Whatever the legal outcome, the lawsuit crystallizes the underlying tension of bot-based cloud transcription: the audio of many people ends up on a vendor's servers, and the rules around consent and reuse are still being worked out. For someone deciding whether Otter is safe for their use, the lawsuit is a signal to take the confidential-meeting question seriously, not a verdict that Otter did something wrong.

The most useful way to read a lawsuit like Brewer v. Otter.ai isn't as a guilty verdict or as noise to dismiss, but as a map of where the risk lives. Lawsuits get filed where there's plausible harm and unsettled law — and AI meeting transcription has both. The harm theory is that people were recorded and their data used to train models without clear consent; the unsettled law is that the rules for AI training on user content are still forming across the whole industry. So regardless of how this specific case resolves, it tells you that the consent-and-training question around cloud meeting bots is a real, contested area — not a solved one. For a casual user that's a footnote. For someone handling confidential meetings, it's a reason to choose an architecture where the question can't arise: if the audio never leaves your device, there's no recording on a vendor's server, no training on your conversations, and nothing for a future lawsuit to be about.

Panel clarifying the Otter.ai Brewer lawsuit is a filed allegation being litigated not a proven verdict but still a signal about confidential meeting privacy

Why Did Universities Ban Otter.ai?

Several major universities — including Cornell, Oxford, and Cambridge — have restricted or blocked AI meeting bots, including Otter, citing privacy concerns. This is worth understanding because these institutions have serious legal and compliance teams, so their decisions are a meaningful signal. The reasoning maps to the architecture rather than to Otter specifically:

A bot in a meeting records everyone present, including students, researchers, and guests
The recording is sent to a third-party vendor's cloud
Universities handle confidential research, student data, and privileged deliberations that shouldn't reach external servers
Obtaining genuine consent from every participant in every meeting is impractical

So the bans aren't "Otter is uniquely dangerous" — they're "the category of bot-plus-cloud meeting tools creates an exposure we can't manage for sensitive conversations." The same logic applies to law firms, hospitals, and any organization with privileged or regulated discussions. It's the architecture being rejected, not one company.

Diagram explaining why universities ban AI meeting bots like Otter because they record everyone send audio to vendor cloud handle confidential data and cannot get consent from all participants

What Does Otter Do With Your Data?

Otter processes and stores meeting audio and transcripts on its cloud servers to provide its features — transcription, speaker identification, summaries, and search. Standard considerations for any cloud transcription service apply:

Audio and transcripts are retained on Otter's servers per its data policies (check current terms for specifics)
Data may be used to improve the service — the AI-training question is central to the lawsuit
Encryption protects data in transit, but the service necessarily decrypts audio to transcribe it, so the vendor has access during processing
Enterprise tiers add controls — SOC 2, SSO, and likely data-handling commitments not in consumer tiers

The thing no cloud transcription service can avoid: to transcribe your meeting, the audio must reach the vendor's servers, where it exists decrypted at least momentarily. Strong security reduces the risk of that data being stolen, but it doesn't change the fact that the vendor has it. For confidential meetings, that's the irreducible exposure — and it's exactly what on-device processing removes.

The AI-training concern at the heart of the lawsuit is worth separating from ordinary security. A service can have excellent security — encryption, SOC 2, no breaches — and still raise privacy concerns if it uses your conversations to train its models, because that's a different question from "can attackers steal my data." Training on user data means your meetings become part of the system that serves everyone, and the consent and control questions there are genuinely unsettled across the whole AI industry, not just for Otter. For most casual users this is an acceptable trade for a free or cheap service. For confidential meetings, it's a reason to prefer tools that can't train on your data because they never receive it — which is the structural property of on-device processing. The audio stays on your machine, so there's nothing to train on and nothing to consent to.

What's the Safer Alternative for Confidential Meetings?

For meetings you can't risk uploading — legal, medical, HR, financial, research, or anything privileged — the safer architecture is on-device transcription, where audio never leaves your Mac. This sidesteps every concern above:

No bot in the call — nothing joins as a participant recording everyone
No cloud upload — audio is processed locally, so there's no vendor server holding it
Nothing to train on — the AI-training concern can't apply to data the vendor never receives
Verifiable — you can confirm it by running the tool offline

Options for on-device meeting transcription on Mac include MetaWhisp (its meeting transcription listens to your computer's audio locally, no bot), Granola (local capture, though summaries go to its cloud), and recording locally then transcribing with MacWhisper. For the broader picture, see our guides on meeting transcription without a bot and private voice-to-text on Mac. This isn't to say everyone should abandon Otter — for non-sensitive meetings its convenience is genuinely useful. It's that the safety of a transcription tool should match the sensitivity of what you're transcribing, and for confidential work, on-device is the architecture that fits.

Diagram matching transcription tool to meeting sensitivity showing cloud tools like Otter for casual meetings and on-device required for legal medical HR and research

A practical rule that resolves most of the "is Otter safe" anxiety: decide based on the most sensitive meeting you'd use the tool for, not the average one. People often reason from their typical meeting — a casual standup where Otter is obviously fine — and conclude the tool is safe, then use it for a confidential call later out of habit. The exposure isn't about most of your meetings; it's about the few that matter. So the right question isn't "are my meetings usually sensitive?" but "will I ever use this for something I couldn't afford to have on a vendor's servers?" If the answer is yes, the safe move is to use an on-device tool for those meetings specifically — or to standardize on on-device so the habit never puts a confidential conversation in the cloud by accident. Matching the tool to your most sensitive use, rather than your typical one, is how you avoid the one upload you'll regret.

Is Otter.ai Safe? The Verdict

For casual, non-sensitive meetings: generally safe — a normal cloud service with enterprise security options.
For confidential or multi-party meetings: proceed with caution — the bot-plus-cloud model is the exposure behind the lawsuit and university bans.
For legal, medical, HR, or privileged work: prefer on-device — match the tool's architecture to the sensitivity of the conversation.
Reality check: the lawsuit is an allegation, not a finding, and the bans reject the category, not just Otter. But both are valid reasons to take confidential-meeting privacy seriously.

Otter isn't uniquely unsafe — it's a capable cloud tool with the trade-offs every cloud meeting bot shares. The question is whether those trade-offs fit what you discuss. For anything confidential, the safest choice keeps the audio on your device.

Frequently Asked Questions

❓

Is Otter.ai safe to use?

For casual, non-sensitive meetings, generally yes — Otter is an established service with enterprise security options (SOC 2, SSO on higher tiers). For confidential meetings, with caveats: Otter's bot joins your call and uploads audio to its cloud. A 2025 class-action lawsuit and bans at several universities reflect concern about this for sensitive conversations. For confidential work, on-device transcription where audio stays on your Mac is safer.

❓

What is the Otter.ai lawsuit about?

In August 2025, a federal class-action lawsuit (Brewer v. Otter.ai) alleged Otter engaged in unauthorized recording of conversations and used user data to train its AI without adequate consent. It's an allegation being litigated, not a proven finding. The core concern is whether meeting participants consented to being recorded by Otter and whether that data trained its models — a question facing AI meeting tools broadly.

❓

Why did universities ban Otter.ai?

Universities including Cornell, Oxford, and Cambridge restricted AI meeting bots, including Otter, citing privacy. The reasoning is architectural: a bot records everyone in a meeting, sends audio to a vendor's cloud, and universities handle confidential research and student data where genuine consent from all participants is impractical. The bans reject the bot-plus-cloud category for sensitive use, not Otter specifically.

❓

Does Otter.ai train AI on my conversations?

The use of user data for AI model training is the central allegation in the 2025 class-action lawsuit (Brewer v. Otter.ai), and is being litigated. Check Otter's current data policies and opt-out settings for specifics. For confidential meetings, the surest way to avoid any training-on-your-data concern is on-device transcription, where the vendor never receives your audio, so there's nothing to train on.

❓

Is Otter.ai HIPAA-compliant?

Otter offers enterprise security (SOC 2, SSO) on higher tiers, but the consumer service is not HIPAA-compliant by default, and a bot recording a medical meeting to the cloud raises clear concerns. For healthcare meetings, on-device transcription where audio never leaves the device is the structurally simpler compliant path — when no third party receives the audio, there's no business associate agreement needed for the transcription step.

❓

What's the safest meeting transcription for confidential work?

On-device transcription where audio never leaves your Mac. MetaWhisp's meeting transcription listens to your computer's audio locally with no bot joining the call. Granola captures locally (summaries go to its cloud). Or record locally and transcribe with MacWhisper. These avoid the bot-plus-cloud exposure behind Otter's lawsuit and the university bans, and you can verify the privacy by running them offline.

About the Author

Andrew Dyuzhov is the solo founder and CEO of MetaWhisp, a free, open-source, on-device voice-to-text app for macOS that runs Whisper large-v3-turbo locally via WhisperKit. He builds an on-device alternative to cloud meeting tools, which is why this assessment discloses that upfront, treats the Otter lawsuit as an allegation rather than a verdict, and draws every claim from open sources — aiming to be a fair privacy review, not a competitor's attack. Connect on X or GitHub.