Bot-Free AI Notetaker Wiretap Risk for Sales Teams

On July 30, 2026, Tarra Chamberlain filed a class action complaint against Granola, Inc. in the U.S. District Court for the Northern District of California, case number 3:26-cv-07926. The complaint alleges Granola’s AI notetaker intercepted meeting communications without obtaining consent from non-using participants. Granola has no visible bot. It captures audio silently through system audio and microphone input. The lead plaintiff never used Granola. She was on a call with someone who did.

Why This Is a Sales Team Problem, Not an IT Problem

Sales managers who switched their teams to Granola specifically because it does not show up as a bot in the meeting made the wrong calculation. The Granola lawsuit targets the invisibility as the mechanism of harm, not the mitigation of it. According to the National Law Review, Granola actively markets the absence of a visible bot as its core feature, with its website stating: “Other people in the room won’t know it’s there.”

That sentence is now the centerpiece of a federal complaint.

There are two separate angles here and they work in opposite directions on any sales call. Both of them land on a sales manager’s desk. Neither has been written for a sales audience yet.

The Take: Bot-Free Made Things Worse, and Your Reps Are Exposed in Both Directions

Every piece of coverage on Chamberlain v. Granola so far addresses this from a compliance, IT-buyer, or in-house counsel angle. None of them address the two things that actually change something on Monday for a sales manager running closers.

First angle: your reps using Granola created the liability, not reduced it. The manager who replaced Otter.ai with Granola to remove meeting friction made the tool’s primary marketing claim into the plaintiffs’ primary exhibit. Invisibility is the offense under the complaint, not a courtesy feature. A visible bot is awkward. An invisible one that captures voice data and uses it to train AI models is the lawsuit. The switch to bot-free did the opposite of what it was supposed to do.

Second angle: your reps are also the non-party on someone else’s call. Holland & Knight flagged this as the novel legal expansion in the Granola filing in their August 2026 alert: Tarra Chamberlain never used Granola. She was recorded by another attendee who had it running. That makes her a non-party plaintiff under the non-party class theory. For your sales team, this runs in reverse on every call your reps take. Your prospects, vendors, and partners may have Granola or a similar bot-free tool running on their side without disclosing it. The same legal theory that creates liability for Granola users makes your reps non-consenting participants in someone else’s recording. They are generating exposure from both sides simultaneously, every day.

The reasonable objection is that these are filed complaints, not decided cases. That is accurate for Chamberlain v. Granola. The related Otter.ai litigation moved faster: Judge Eumi K. Lee issued an order on August 13, 2026, allowing claims to proceed under federal wiretap law, California privacy law, and Illinois biometric law. The theory survived a motion to dismiss. Chamberlain is earlier in that same lifecycle. But in neither case does “waiting for a ruling” constitute a consent process, and that consent gap is what this article is actually about.

The Evidence

Three lawsuits, a pattern. Chamberlain v. Granola, No. 3:26-cv-07926 (N.D. Cal., filed July 30, 2026) joins In re Otter.AI Privacy Litigation (N.D. Cal.) and Fireflies.ai suits in Illinois and the Northern District of California. Granola and Otter.ai are being sued under the Electronic Communications Privacy Act (ECPA) and state wiretapping statutes including California’s Invasion of Privacy Act (CIPA). The Fireflies.ai complaints run separate claims under the Illinois Biometric Information Privacy Act (BIPA), targeting voiceprint capture. Different legal theories, same pattern: AI notetakers capturing conversations without adequate consent. The National Law Review has the full case summary at natlawreview.com.

Federal one-party consent is not the floor in all-party consent states. Under the ECPA, only one party to a call needs to consent to the recording, meaning the Granola user’s own consent is all federal law requires. Non-participating attendees receive no notification and cannot meaningfully withhold consent, per Mondaq’s analysis. State law is where this breaks. California and Illinois, the two states where these suits are concentrated, both require all-party consent. California’s CIPA provides statutory damages of $5,000 per violation per participant, per Basil AI’s complaint analysis at basilai.app. If your team closes calls into either of those states, federal one-party consent offers no protection.

The non-party class theory is the move nobody planned for. Holland & Knight’s alert identifies this as the novel element in the Granola case. Chamberlain was recorded by another attendee’s Granola session without any knowledge it was happening. That creates a plaintiff class that extends beyond tool users to anyone who was on a call where the tool ran. Your reps are in that class on any call where the other side has a bot-free recorder running silently.

The training data layer adds another dimension. The Granola complaint alleges the tool defaults to using captured audio for AI model training, and that once data is incorporated into models it cannot be extracted, per the National Law Review’s reporting. If your reps’ voice data is being ingested into a third-party AI model without their knowledge, the consent problem is not only about the transcript.

What to Do This Week

1. Find out what tools your reps are actually running on calls. You know what the company-issued stack looks like. You may not know what reps have installed on personal accounts. Granola is a personal subscription under $25 a month. Ask directly, this week, not in the next all-hands.
2. Add a verbal disclosure to your call opener. One sentence: “I’m using an AI notetaker on this call.” That is not a legal opinion. It is a consent signal. If the prospect continues the call after that disclosure, you have documented conduct consistent with consent. Build it into the call opener script your reps rehearse, the same place the intro and agenda live. A policy document no one reads is not a disclosure process.
3. Know your call volume into California and Illinois. Those are the states where these lawsuits are running under all-party consent statutes. If a significant share of your closing calls flows into either state, that is where your exposure concentrates. Know the number before a legal context forces the question.
4. Check the default training data setting on your current tool. Per the Granola complaint, the tool defaults to using captured audio for model training. Whether your reps know their tool’s default is a question worth answering now. Discovery is a bad time to find out.
5. Do not assume switching tools closes this gap. Otter.ai, Fireflies.ai, and Granola are all being sued under the same legal theory. The tool is not the variable. The missing consent step is. A team that moves to a new recorder without adding a verbal disclosure to its opener has not fixed anything. It has just changed the brand name on the exposure.

The Real Gap

All three of these lawsuits share the same structure. Recording happened silently, at scale, in a context where the recorded party had no practical way to notice it and no moment in the process where they could object. Sales calls are the highest-frequency version of that context in business. The closer wants the tool. The prospect is focused on the pitch. The consent moment never happens because nobody built it into the protocol, because for most of the history of call recording the bot was visible enough that people understood something was capturing the conversation. That is no longer the situation.

The fix is not a different tool. It is a different call opener, built for a world where recording is invisible by default, so the consent step becomes the rep’s responsibility rather than the software’s.

eNZeTi is built for sales managers who coach by reviewing calls without listening to every one. If your team is going to run a call review program at scale, the consent disclosure needs to exist before the recording starts. That is what makes the program defensible, not just useful, and it is the operating assumption eNZeTi is built on.