Zoom Says It Isn’t Training AI On Private Calls Without Consent, But It Might Be Using Other Data

Zoom, a well-known videoconferencing platform, has faced scrutiny due to updates in its terms of service that seemed to grant the company permission to utilize user data, including private conversations, to train AI models. This claim gained traction on social media, with one tweet asserting that Zoom’s terms now demand unconditional and irrevocable consent for AI training, leaving users with no opt-out option.

Responding to these concerns, Zoom issued a blog post to clarify its position. The post emphasized that the company “will not use audio, video, or chat customer content to train our artificial intelligence models without your consent.” The company also added language to the terms of service to reinforce this point.

Privacy experts acknowledged that the revised terms accurately reflect the company’s intent. While user-generated content like video and chat cannot be used for AI training without user consent, participants in meetings hosted by someone who agrees to AI training could potentially be required to leave if they disagree with the data usage.

Zoom’s terms allow the company to utilize other data, such as information about user behavior, for AI training without explicit consent. However, experts raised concerns about the extent of control participants have over their data when the host or administrator consents to data usage.

The issue’s core revolves around language added to Zoom’s terms of service in March, distinguishing between “service generated data” and “customer content.” The former includes data about users’ feature usage and geographical location, while the latter involves user content, like audio or chat transcripts. The terms initially suggested that service-generated data could be used for AI training, but the company has now clarified that customer consent is required for such usage.

Despite the clarification, experts remain cautious about the practical implications. Zoom’s approach to consent, especially in scenarios where a host enables generative AI features, could limit users’ ability to control their data usage. For instance, participants might not opt out of data sharing if their boss mandates using Zoom for meetings.

In essence, while Zoom has taken steps to address the concerns about AI training and user consent, the intricacies of its features and consent processes have prompted ongoing discussions about the true extent of user control and privacy in the platform’s operations.

Leave a Reply

Your email address will not be published. Required fields are marked *