Zoom, the popular videoconferencing platform, has revised its Terms of Service in response to a recent uproar over the company’s policies regarding the training of artificial intelligence (AI) using user data. The initial policy explicitly granted Zoom the right to train AI models on user calls without requiring explicit user permission. However, an added clause now assures users that Zoom commits not to employ audio, video, or chat Customer Content for AI model training without their consent.
The revised terms, introduced last August 7th, come in the wake of a contentious debate triggered by a post on the Hacker News forum over the weekend. The language alterations within Zoom’s legal documentation now include a bold statement clarifying the company’s commitment to obtaining user consent before utilizing customer content for AI model development.
Zoom’s AI policies, which had previously attracted minimal attention, have been under scrutiny recently. The company’s Terms of Service categorizes users’ video, audio, and chat transcripts as “Customer Content.” By accepting these terms, users grant Zoom extensive rights to their data for purposes such as machine learning, AI training, and product development. A similar scope of rights is reserved for “Service Generated Data,” encompassing telemetry, usage, and diagnostic data obtained through content and behavior analysis.
According to Smita Hashim, Chief Product Officer at Zoom, AI is currently being employed in various ways, including the Zoom IQ Meeting Summary feature, which automatically generates meeting summaries. The company also utilizes AI for tasks like scanning webinar invitations to identify spam activity. In a blog post, Hashim emphasized that meeting administrators have the choice to decline sharing meeting summaries data with Zoom. Additionally, non-administrator attendees are informed about the new data-sharing policies and given the choice to either accept or exit meetings.
Responding to concerns, a Zoom spokesperson affirmed that customer consent and control remain pivotal. Zoom users have the discretion to activate generative AI features and to determine whether their customer content is shared with the platform for the purpose of enhancing products. Despite these clarifications, data privacy advocates and some users remain apprehensive, urging further revisions to the language and expanded user options to opt out of AI training data usage.
