A few years ago, AI note-taking and transcription tools were a niche productivity hack used by busy executives and the media. Now, they're everywhere. But these apps come with a legal fine print that most people don’t know to look for.

First, the app may retain recordings and transcripts of your meetings indefinitely once used. They can also reuse them for any number of internal purposes, including model training and marketing campaigns. Second, state laws often require all parties to consent before a conversation is recorded. AI note-takers argue that the onus of obtaining consent falls on the person using the recording app, aka you. So if you recorded a meeting but didn’t disclose it to the other party, they could actually sue you for violating their data protection rights.

That’s why multiple AI note-taking platforms are now facing class-action lawsuits from users who accuse them of mishandling their data and passing the baton of consent down to customers without informing them about this properly. I'll walk through the current legal situation, as well as what steps you can take to protect your data going forward.

What the Granola lawsuit is all about

Chamberlain v. Granola, Inc. was filed on July 30, 2026, in a California district court by plaintiff Tarra Chamberlain, who says that she was recorded without her consent (or even knowledge) while on a video call with someone who was using Granola. Unlike most AI note-takers, which show up as bots that have to be visibly admitted into the video call, Granola is built around invisibility. According to the app’s own marketing copy, the app's competitive advantage is it runs locally on a user’s device so that “other people in the room won't know it's there.” Lawyers fighting for the plaintiff say that this is the textbook definition of an illegal wiretap.

Moreover, by design, no one in the meeting can turn Granola off, except for the user who has the app installed on their device. Users are also opted into model training by default, so they must manually opt out if they don’t want the company to retain the conversation records. Even if a user does opt out of model training, this setting does not apply backwards, and prior conversation records are still retained for training purposes.

In their formal complaint, the plaintiffs’ lawyers cite multiple federal and state law violations, including the Electronic Communications Privacy Act and California's Invasion of Privacy Act. Granola denies these accusations by saying that it anonymizes all training data before use and that no conversation data is sent to third parties, but the plaintiffs reject this framing.

Granola isn’t the only one with the problem

This isn’t the only active class action against an AI note taker in Silicon Valley. In 2025, four class action lawsuits against Otter.ai were consolidated into a single case before Judge Eumi K. Lee in the Northern District of California. Like Granola, this case alleges Otter records conversations without consent from meeting participants and then uses the recordings to train their own AI models. Justin Brewer, one of the plaintiffs in the suit against Otter, says that he was recorded using the OtterPilot app during a sales call and his voice was subsequently used to train AI models without his knowledge. While the lawyers working for Otter have filed a motion to dismiss, there hasn’t been a ruling yet. 

Fireflies.ai is also facing multiple class action lawsuits for violating the Illinois Biometric Information Privacy Act (BIPA) by harvesting voiceprints from meeting participants using its speaker recognition technology without obtaining consent from those recorded. Read AI has not faced a lawsuit, but the app is banned from multiple public and private universities, including the University of Washington, Chapman University, Tufts University, and Mississippi State University. A notice on the University of Washington website reads, “Read AI can join, transcribe, and summarize their users’ online meetings even when the users are not in attendance,” adding that it poses significant “security and privacy risks to institutional data.”

Who is responsible for obtaining consent with AI note-taking apps?

In every one of these class action suits so far, AI note-takers insist that the onus of obtaining consent from other participants falls upon the user, not the company that built the app. Otter.ai has specifically stated that this is baked into the contract that new users are forced to sign, which outsources the company’s consent obligations to the user. So if you used a note-taker to record the call, Otter.ai says it’s your job to make sure your fellow participants are aware, and they all consent to the recording.

Of course, all of this is based on a clause in the user agreement you sign when creating a new account with the note-taking app, which is no doubt buried under heaps of other text that no one bothers to read. Historically, courts don’t look favorably upon tech companies hiding behind terms of service agreements and passing accountability onto users. But that’s not granted and still subject to case-by-case determination by the presiding judge.

Then again, all-party consent isn’t a legal requirement for call recording in every U.S. state. But at least eleven of them, including California, Florida, Illinois, Maryland, Massachusetts, Montana, New Hampshire, Pennsylvania, and Washington, enforce similar rules to protect user privacy. Depending on the circumstances of the call and information exposed, other states might also offer certain protections. Given the confusing circumstances and the lack of a ruling on any of these lawsuits, you shouldn’t assume that you’re not responsible for obtaining consent because you use a third-party app or service. But that doesn’t mean you should let AI note takers skirt blame entirely either. 

How to protect your individual rights with AI note-taking apps

There’s no denying the convenience of AI transcriptions, plus you can’t really control what software other people in a meeting choose to run in secret. But there are some things you can do to protect yourself from privacy violations and compliance issues, both as a user and another meeting participant.

• If you choose to run an AI note-taker during a meeting, - To that point, if anyone in the meeting is a resident of California, Illinois, or another all-party consent state, assume that you must verbally inform and ask permission before you begin recording.
• If you are joining a meeting, ask the meeting host directly if the conversation is being recorded or if a note-taker is being used, if they did not disclose that fact ahead of time.
• Before you give an AI app access to your meetings, make sure that you’ve opted out of any model training or data sharing agreements in your app settings.
• Depending on your region of residence, you may be eligible to legally revoke permissions to access your past data using a written notice. For example, California residents can invoke their right to erasure when asking a company to delete personal data. If so, contact the app's customer service line.

None of this can fully guarantee that you’ll never be subjected to privacy violations from an AI meeting bot, but being aware and conscious helps you stand a better chance of defending your rights. If a meeting requires disclosing sensitive information and you’re unsure if you’re being recorded by an AI app, it’s best to leave the meeting entirely and obtain written confirmation from the host before proceeding.