NEW YORK (AP) — The idea of letting an artificial intelligence tool handle all your meeting notes sounds appealing. Within moments of sitting in on an hour-long video call, these AI agents can produce a full recap of what was discussed and generate a task list for everyone involved.
But the way these tools actually work has led many professionals to think twice before using them. At their core, AI notetakers convert everything spoken during a meeting into data — and that data can include confidential personnel matters, sensitive business strategies, trade secrets, and potentially damaging remarks that could fall into the wrong hands.
“There are huge risks to the organization on AI notetakers,” said Amy Dufrane, the chief executive of human resources training and certification provider HRCI. “I don’t think companies should use it at all.”
An AI notetaker is a software program or device that combines artificial intelligence, voice recognition, and large language models to capture, transcribe, and condense conversations. While the tools are designed to save time and boost engagement, professionals across multiple industries say there are legitimate reasons for concern.
One of the biggest worries is not knowing where recorded data ends up or how long it’s kept. Privacy advocates are raising alarms that the companies behind these tools may be building voiceprints — unique acoustic profiles similar to fingerprints but based on the specific qualities of a person’s voice — without obtaining consent. Those voiceprints can be used to gain access to restricted accounts, including bank accounts.
Some technology companies resell data gathered through their notetaking products or use recordings and transcripts from private meetings to improve their AI systems. There’s also a legal dimension: a federal judge in New York ruled in February that a criminal defendant had to hand over documents he had prepared for his attorneys, because those documents had already been shared with a third party — Anthropic’s Claude AI.
“People who use AI notetakers, they don’t always know where the data goes,” said Justin Daniels, an Atlanta-based corporate attorney at law firm Baker Donelson. “And in my context, if the data goes anywhere else and they’re not aware of it, that attorney-client-privileged conversation may not be attorney-client-privileged anymore.”
Here is what you should know about the risks of AI notetakers, how to handle them in a meeting, and steps you can take to protect your information.
When you enter a virtual meeting, get into the habit of checking for an AI notetaker. It may appear on the participant list — often with a label indicating it’s an AI tool — or a notification may pop up on screen alerting attendees that the session is being recorded, which can indicate a notetaker is active.
Platforms like Zoom and Google Meet typically alert users when recording is happening, but not all meeting software makes it obvious when an AI notetaker has been added, according to Thorin Klosowski, a senior security and privacy analyst at the Electronic Frontier Foundation.
It’s also possible for participants to use a separate personal device to record and transcribe a conversation without the knowledge of others in the meeting.
“You hope the other person would tell you that they’re doing that,” Klosowski said. “Asking everyone for consent before doing a sensitive meeting would be the most polite approach to take.”
If you’re not certain whether an AI notetaker is present, simply ask. You can also set expectations at the start by stating the meeting is not authorized to be recorded.
Dufrane suggested a straightforward way to handle it: say, “Our company policy is that this meeting cannot be recorded.” Framing it as a company rule takes the pressure off individual employees — like a salesperson trying to make a good impression — and shifts responsibility to the organization, she explained.
Another approach is to allow the notetaker to run during part of the meeting, then shut it off before turning to more sensitive topics.
“I won’t start talking about anything substantive until it’s shut off, because I just don’t want to take the risk,” Daniels said.
According to Chris Pluymers, an associate attorney at The Dillon Law Group in East Lansing, Michigan, many AI notetaking tools create individual voiceprints for each person speaking — that’s how the software tells participants apart and labels them as “Speaker 1” or “Speaker 2.”
Voiceprints are already used by some institutions to verify the identity of bank customers over the phone. If those vocal signatures were obtained by bad actors, they could be weaponized to access files, commit fraud, or take over accounts, Pluymers said.
Several states have enacted laws governing how voiceprints can be collected and stored, and those laws give individuals certain rights to object to the use of AI notetakers in meetings they attend.
In Illinois, voiceprints are classified as biometric identifiers — similar to fingerprints — and fall under the state’s Biometric Information Privacy Act. That law requires written notice and informed consent before any AI tool collects voiceprints. It also requires companies to have a documented plan for how long data is kept and when it will be destroyed. However, Pluymers noted that most companies using these tools don’t have any of those safeguards in place.
“In the world of AI, the world of data and privacy, the world of biometric identification, I don’t think you can have such a lax approach to it,” Pluymers said. “I think getting out ahead of it is crucial.”
Under that Illinois law, workers can decline to attend a meeting where an AI notetaker is present until they receive clear answers about how their data will be stored, why it’s being kept, and when it will be deleted, Pluymers said. They can also request to see any written policy or consent form related to the tool.
If an AI notetaker shows up unexpectedly, Pluymers suggested a polite response: “I prefer we keep this meeting without AI recording or transcript tools and I’d be happy to take my own notes and share a recap if that’s helpful.” Being respectful while clearly expressing your preference is often the most effective approach, he said.
Danielle Kays, a partner at Fisher Phillips who advises businesses on privacy and employment law, recommends finding out whether the companies behind AI notetaking apps hold on to recordings, transcripts, or metadata indefinitely — or use that material to train their AI systems.
“If there is some sort of speaker ID or voice recognition, really understand what that is and how it works,” Kays said.
Even when content appears to be deleted, metadata about meetings can remain stored with the vendor. That means sensitive business details could still influence how an AI model behaves and, in some cases, could be recalled or reproduced by the system, she said.
The Electronic Frontier Foundation also points out that AI notetakers produce text-based output, which is far easier for outside parties to search through than raw video or audio files.
“Storing a bunch of video isn’t easy, it’s costly and hard to look through, but text is much easier to search and cheaper to store,” said Klosowski of the Electronic Frontier Foundation.
