Are AI Tools Listening In on Your Meetings?

In today’s digital landscape, it’s becoming increasingly common for meetings to be recorded not just for the sake of documentation, but also through advanced artificial intelligence tools. These tools have revolutionized how many professionals conduct their meetings, but they come with significant privacy implications that deserve scrutiny.

The Rise of AI Meeting Tools

Artificial intelligence-based applications such as Fireflies.ai, Otter.ai, Trint, and Fathom are gaining traction among professionals, from corporate boards to healthcare practitioners. These tools simplify the process of recording, transcribing, and summarizing meetings. While some users are fully aware of their presence—seeing a designated participant in a video conferencing session or a visible recording indicator—others run these tools silently in the background.

According to Nicolas Joubert, a legal expert from the law firm MLT Aikins in Winnipeg, the increase in AI tool usage is creating a complex privacy landscape. “You can hardly go a week without a news story highlighting an AI mishap or a data breach,” Joubert notes.

Advantages of AI Notetakers

Despite the associated risks, many professionals appreciate the benefits of AI transcription tools, which help to reduce the monotony of meetings and enable attendees to engage more fully in discussions. Kael Campbell, president of Red Seal Recruiting Solutions Ltd., shared that his hiring firm has been using HoneIt for about four years. He finds the transcripts to be notably comprehensive, allowing him to provide clients with specific information when needed. “Previously, I wouldn’t take full verbatim notes, but now I can reference full transcripts,” he said.

Risks Involved

However, AI tools also come with a host of challenges. They tend to generate large volumes of data, which may contain errors and personal information, presenting various privacy concerns. Joubert highlights that such tools often record both casual chatter and essential business discussions, inadvertently capturing sensitive personal information.

These tools also struggle with distinguishing moments when discussions are meant to be confidential. An in-camera meeting could inadvertently be recorded, compromising privacy. Teresa Scassa, Canada Research Chair in Information Law and Policy at the University of Ottawa, stressed the potential fallout: “Suddenly, those private discussions are shared with an entire mailing list, which poses significant risks.”

Accuracy and Data Vulnerability

In addition to privacy concerns, the accuracy of transcriptions can be questionable. Technical issues or unfamiliar terminology may lead to errors, sometimes referred to as “hallucinations,” where the AI fabricates incorrect details. Furthermore, the data collected can be vulnerable to breaches, as many recordings and transcriptions reside in cloud storage. Joubert emphasizes the importance of examining the terms of service associated with these tools, questioning where data is stored, how it is processed, and whether it’s being used to train new AI models.

“These questions are critical to address before integrating these systems into your workflow,” Joubert cautions.

Real-World Consequences

There have been notable incidents where participants were unaware of recordings, often learning only when their information was mistakenly distributed. Scassa cites an example from September 2024, where an Ontario hospital faced a serious breach when a doctor’s use of Otter.ai resulted in the recording of patient discussions without notice. The recordings were later sent to all invitees, forcing the hospital to report the incident and revise its protocols.

Best Practices for AI Tool Usage

Experts like Joubert and Scassa advocate for transparency when using AI tools in meetings. They recommend notifying participants and considering their concerns. While Red Seal’s clients generally accept AI usage, some prefer that notes be deleted afterward. Campbell emphasized, “If anyone objects, we revert to manual note-taking, despite the extra time involved.” His team ensures that all meeting participants are informed about the use of any AI tool and double-checks the accuracy of outputs before sharing them.

Understanding Risk and Responsibility

Before adopting such software, businesses should closely review the terms to understand the scope of data rights they are granting. Joubert pointed out that many larger vendors tend to minimize their own liability while placing more onus on the customer. He acknowledges the utility of AI meeting tools but warns against the risks of allowing sensitive information to be accessed by unintended parties.

“You wouldn’t want to broadcast your private information on a busy street,” he explained. “Using these tools is no different.”

Conclusion

As AI tools continue to change how we conduct meetings, it’s crucial to balance their advantages with the potential risks they pose to privacy and data security. Understanding and managing these risks can help professionals leverage the benefits of AI while protecting sensitive information.