AI Note-Taker Lawsuits: The Pattern

A wave of class actions accuses AI note-takers of recording private meetings without everyone's consent. The suits against Granola, Otter.ai, and Fireflies all lean on the same federal wiretap theory and the same fight over consent.

These cases are early. Every claim described here is an allegation. No court has found that any of these tools broke the law.

This is general information, not legal advice. If your team records meetings, consult qualified counsel about your specific situation.

What Do These Lawsuits Have in Common?

All three suits allege the same thing: an AI note-taker captured private conversations and used them without the consent of every participant. Granola, Otter.ai, and Fireflies each face a proposed privacy class action built on that claim.

The lead legal theory is identical across them. Each casts the note-taker as an uninvited listener under the federal wiretap statute, then adds state consent claims on top.

Because the theory is shared, a ruling in one case can shape the others. That is why the whole industry is watching a single pending decision rather than three separate ones.

The products are not identical, and those differences matter. Some send a visible assistant into the call, while others capture quietly in the background, and courts may treat a tool that announces itself differently from one that never appears. But the legal spine, an interception without full consent, is the same in every complaint.

The Shared ECPA Wiretap Theory

The common thread is the Electronic Communications Privacy Act, the modern Wiretap Act. It bars intentionally intercepting the contents of an electronic communication without authorization.

Plaintiffs argue that an AI note-taker intercepts conversation contents in real time, which can bring it inside the statute. The vendors respond that the user who turned the tool on consented, and that one party's consent is enough under federal law.

The plaintiffs' escape hatch is the 'criminal or tortious purpose' exception, which removes the consent defense when interception serves an unlawful aim. The Ninth Circuit reads it narrowly, so how each court applies it is decisive.

District courts in California have not lined up cleanly on that exception. Some read it to strip consent whenever the interception itself is a tort; others require proof that an unlawful purpose was the reason for the interception. Until a higher court settles the split, similar facts can produce different results depending on the judge.

That uncertainty is the engine behind the whole wave of suits. Plaintiffs' firms file where the law is unsettled, because an unsettled question is one a court might answer in their favor.

The Three Cases at a Glance

Granola faces Chamberlain v. Granola, Inc., No. 3:26-cv-07926, filed July 30, 2026 in the Northern District of California. The complaint focuses on Granola's silent, no-bot capture model and alleges the tool trains AI on captured conversations by default.

Otter.ai faces a consolidated privacy class action in the same court, where a closely watched ruling is expected on whether the service is a tool of its user or a separate eavesdropper. Fireflies faces a parallel suit on the same wiretap theory.

Different products, one legal spine. The factual details vary, but the consent question sits at the center of all three.

• Granola: Chamberlain v. Granola, Inc. (N.D. Cal.), silent no-bot capture, alleged training-by-default.
• Otter.ai: consolidated N.D. Cal. class action, pending ruling on the tool-versus-eavesdropper question.
• Fireflies: parallel class action on the same ECPA wiretap theory.

Recurring Issue One: Non-consenting Participants

The first recurring issue is the person who never agreed to be recorded. AI note-takers are usually turned on by one participant, but a meeting has many, and the others may not know capture is happening.

Silent tools sharpen the problem. When nothing appears in the participant list, other people on the call get no notice at all, so their consent is never sought.

In our AI workflow audits for teams rolling out these tools, the most common gap is a consent step that exists on paper but never happens in the actual meeting.

Recurring Issue Two: Training-data Defaults

The second recurring issue is what happens to the recording after the meeting. Several complaints allege the vendor uses captured conversations to train AI models by default, not only when a user opts in.

That raises a permanence concern. Plaintiffs argue that once conversation data is absorbed into a trained model, it cannot be cleanly pulled back out, so a later opt-out does not fully undo it.

For a deploying company, the practical takeaway is simple. Check the training setting before rollout, because the default may send sensitive client conversations into a model you cannot claw back.

Recurring Issue Three: One-party vs. All-party Consent

The third recurring issue is which consent rule applies. Federal law and most states allow one-party consent, while about a dozen states require all-party (two-party) consent.

California's CIPA is the leading all-party statute in these cases, and it is harder for a silent note-taker to satisfy. Because state laws differ so much, nationwide class certification becomes a real fight, and the plaintiff's home state shapes which claims can lead.

For deployers, the map matters more than the vendor's lawsuit. Recording a call that touches an all-party state can create exposure no matter how the litigation ends.

Remote work makes this messier. Participants on one call can sit in several states at once, and courts often apply the stricter rule when even one participant is covered by an all-party statute. A team that treats every external call as all-party avoids having to guess where each person is sitting.

Read the Individual Case Breakdowns

Each case has its own facts worth understanding in detail. The Granola breakdown covers the silent-capture model and the training-by-default allegation. The Otter.ai breakdown covers the pending ruling and the tool-versus-eavesdropper question.

For the consent rules themselves, start with the guides on two-party consent states and whether it is illegal to record a conversation. Then use the consent-compliance guide to turn the rules into a meeting process.

The links below connect the case coverage with the consent fundamentals, so you can move from what is happening in court to what to change in your own workflow.

• Granola case detail: see The Granola Lawsuit, Explained.
• Otter.ai case detail: see The Otter.ai Lawsuit, Explained.
• Consent rules: see Two-Party Consent States and Is It Illegal to Record a Conversation?
• Build a process: see AI Note-Taker Consent Compliance.

What Should Deployers Watch and Do?

Watch the pending Otter.ai ruling first. If a court decides an AI note-taker is a third-party eavesdropper, exposure rises across every tool built the same way.

In the meantime, fix the process rather than the tool. Announce recording, get consent on the record, and confirm training settings, with extra care on external calls in all-party states.

Treat vendor choice as a governance decision, not just a features decision. In our governance reviews, the teams that avoid trouble are the ones that decided how they would get consent before they decided which note-taker to buy.

None of this requires abandoning AI note-takers. It requires using them on purpose, with a consent step and a training decision made once and applied to every meeting, instead of leaving each employee to improvise.

• Add a spoken recording disclosure to every external meeting.
• Get consent on the record in all-party consent states.
• Audit each tool's default training and data-retention settings before rollout.
• Assign an owner for note-taker governance instead of leaving it to individual users.