Otter.ai Lawsuit: The Ruling Every AI Notetaker Should Read

The short answer: On 13 August 2026 a federal judge in California let most of the privacy claims against Otter.ai move forward. She held that a notetaker vendor can be treated as a third-party eavesdropper if it independently collects, retains and uses recordings for its own commercial purposes. It is a pleading-stage ruling, not a finding that Otter broke the law. For the case’s current status, see our AI notetaker lawsuits tracker.

This post is a dated record of the August ruling. We have corrected it against the court’s order, but we have not rewritten the analysis.

What was decided

Judge Eumi K. Lee ruled on Otter’s motion to dismiss on 13 August 2026. She rejected Otter’s argument that the plaintiffs lacked standing and denied the motion on most of the claims. These survived:

• the federal Electronic Communications Privacy Act (ECPA);
• California’s Invasion of Privacy Act (CIPA), section 631, for all three California plaintiffs, and section 632 for one of them;
• the Illinois Biometric Information Privacy Act (BIPA);
• unjust enrichment, California’s Unfair Competition Law (UCL) and a declaratory-judgment claim;
• the intrusion-upon-seclusion claim of one California plaintiff.

These were dismissed, with leave to amend: the Computer Fraud and Abuse Act (CFAA), California’s Comprehensive Computer Data Access and Fraud Act (CDAFA), the Washington Privacy Act claim, and the common-law privacy claims of the other plaintiffs. Plaintiffs also withdrew several claims, including conversion, trespass to chattels and larceny.

Two things to be clear about before going further. This is a pleading-stage ruling. It decides that the plaintiffs alleged enough to proceed to discovery, treating their allegations as true. It does not decide that Otter broke the law, and Otter denies the allegations.

The facts that got it here are worth recalling, because they are so ordinary. As alleged, one plaintiff, Justin Brewer, joined a Zoom meeting where Otter’s notetaker was running, with no account and no opportunity to decline. The complaint says the notetaker joins as a “silent participant,” records and transcribes in real time, and keeps the content for Otter’s own use, including model training.

That is not an exotic scenario. That is Tuesday.

The holding that generalises

Here is the part vendors should be reading closely.

California’s CIPA section 631 targets, among other things, third parties who eavesdrop on a communication. The natural defence for a software vendor is that it is not a third party at all. It is a tool, an extension of the customer who deployed it, in the same way a telephone or a notepad is. Courts have accepted versions of that argument before.

The court did not accept it here. Because the plaintiffs plausibly allege that Otter “independently collects, retains, and uses communications for its own commercial purposes,” it held they had sufficiently alleged that Otter is a third-party eavesdropper under section 631.

Read that again with your own vendor in mind. The distinction the court drew is not about bots. It is not about notification. It is not about California. It is about whether the vendor does something with the recording for itself.

Training a model on customer conversations is exactly that. So is retaining recordings indefinitely to improve a product. A tool that transcribes and hands the result back looks like a tool. A tool that keeps a copy and mines it looks, on this reasoning, like a participant with its own interests. Whether that reasoning survives later stages is for the court to decide.

Where the major tools stand

Of the 28 AI meeting assistants we rank, several train on customer data by default unless the user finds and flips a setting. Otter.ai does: its privacy policy, effective 16 June 2026, lists training its proprietary AI on de-identified audio recordings among its purposes. So do Granola and Fathom, which train on de-identified data unless you switch it off in settings. Notta is unclear: it lists “no AI training” as an Enterprise feature, which implies lower tiers may be used. Our methodology scores data retention and training transparency at double weight for exactly this reason.

The tools that look best after this ruling made the opposite architectural choice. Circleback and Jamie both state plainly that they do not train on customer data, and Jamie deletes meeting audio after transcription. Fireflies.ai states it does not train on meeting content. Those are contractual and procedural positions, and they are the right ones for a cloud tool.

The structural answer is not to hold a copy at all. Hedy runs speech recognition on your device, and with Local AI on supported hardware (shipped May 2026) the summaries, notes and live coaching run there too. In that setup the recording never has to reach a Hedy server, so there is no vendor-held copy to collect, retain or use. If you choose Hedy’s cloud AI instead, the transcript goes to Hedy’s EU or US cloud, where its AI providers are contractually barred from training on it. Hedy never trains on your data in either mode.

That is a structurally different exposure from a promise not to look. It is not a legal shield, and we say so plainly below.

The other thing vendors should notice

Bot-free capture is having a moment, and this ruling complicates the marketing around it in a way worth naming.

Recording without a bot is not itself a problem. Several tools we rate highly capture from the user’s own device, and doing so avoids the meeting-platform friction that has made bots increasingly unwelcome. But the absence of a bot means the absence of an announcement, and the absence of an announcement means the other participants may not know.

Some vendors handle this responsibly, pairing bot-free capture with clear guidance that the user must disclose. Bluedot, by contrast, leads with its invisibility to other participants. After a ruling that turned on non-consensual capture of someone who had never agreed to anything, that positioning looks less clever than it did in July.

The test case arrived two weeks before this ruling. On 30 July, a Florida plaintiff who had never signed up for Granola filed Chamberlain v. Granola, Inc. in the same district, alleging that another participant’s silent capture recorded her on a Teams or Zoom call and that Granola used the content to train models by default. That filing is at the pleading stage, and we cover it separately in Granola’s lawsuit. The point for this piece is simple: the Otter holding was never a bot holding. Granola is what it looks like when plaintiffs apply the same theory to a tool that never joins the call.

Bot-free is not consent-free, including with Hedy. Hedy captures system audio without a bot, and it keeps recordings off a vendor server when you use on-device processing. Neither fact removes your duty to tell the other people on the call. Say out loud at the start that you are recording, and note the agreement. In all-party consent states such as California, Illinois and Pennsylvania, that is not etiquette, it is the law. Under GDPR, one participant’s consent has never covered the rest of the room.

What to do this quarter

Find out whether your assistant trains on your conversations, and whether that is the default. Not what the marketing page implies. What the data processing terms say. If it is opt-out, someone in your organisation has almost certainly not opted out. Our guide to opting out goes tool by tool.

Check your retention setting. Indefinite retention is the second half of the “collects, retains, and uses” formula.

Write down your disclosure practice. If you use bot-free capture, the obligation to tell people sits with your staff, and staff need to know that.

Ask about Illinois. BIPA survived here, and BIPA has a private right of action with statutory damages. Any tool doing voiceprint-based speaker identification on Illinois residents deserves a specific conversation.

Shortlist by architecture. Our privacy guide and guide for lawyers rank tools by where the audio goes and whether the vendor trains on it. If you are leaving Otter, the Hedy comparison shows the trade-offs.

The direction of travel

The EU AI Act’s general application began on 2 August 2026, eleven days before this ruling. Its workplace prohibition on inferring emotions from biometric data has applied since February 2025, and consent cannot cure it. A US court has now held, at the pleading stage, that what a vendor does with a recording can determine whether it was eavesdropping in the first place.

These are different legal systems reaching for the same idea from opposite ends. Both are moving away from “did someone click accept” and toward “what does this software actually do with what it hears.” That is a harder question to paper over, and it rewards products designed to need less.

Update, 27 October 2026: Otter filed its answer on 17 September and the class-certification schedule runs into 2027. See the AI notetaker lawsuits tracker for the latest. Source for the ruling: the court’s order of 13 August 2026 in the CourtListener docket for No. 5:25-cv-06911 (Dkt. 68).

This is analysis, not legal advice. If your organisation records meetings across multiple jurisdictions, take advice specific to your deployment.