Zoom revises terms again to say it doesn’t use customer data to train AI models
Zoom has updated its terms of service (TOS) again to remove language about using content collected from its communication app to train artificial intelligence models, following a barrage of criticism from users and privacy advocates.
“Following feedback received regarding Zoom’s recently updated terms of service Zoom has updated our terms of service and the below blog post to make it clear that Zoom does not use any of your audio, video, chat, screen sharing, attachments, or other communications like customer content (such as poll results, whiteboard, and reactions) to train Zoom’s or third-party artificial intelligence models,” the company said in an updated blog post on Friday.
The move was a further rollback of changes made earlier in the week, when Zoom clarified that it would first obtain customers’ consent to use their data to train AI models. The technology website Stack Diary reported on August 6 that Zoom’s TOS had been changed in March to say the company had a "perpetual, worldwide, non-exclusive, royalty-free, sublicensable, and transferable license" to use the data any way it wants, including for training AI models.
The reporting prompted a wave of pushback on social media, with some arguing that the TOS could run afoul of the EU’s General Data Protection Regulation (GDPR) due to the way it obtains consent from users.
The criticism — and subsequent rollback — will likely be a cautionary tale for technology companies that are trying to find ways to develop lucrative AI tools. AI innovation is forecasted to boost U.S. stocks, according to Goldman Sachs, with companies like Nvidia — which produces advanced chips used in AI systems — seeing their stock price nearly triple in 2023.
Zoom recently introduced two generative AI features, Zoom IQ Meeting Summary and Zoom IQ Team Chat Compose, which create automated meeting summaries and AI-powered chat composition. Although the company initially said it would potentially use customer data to enhance these tools, it now says that it does not use any communication data “to train Zoom’s or third-party artificial intelligence models.”
Additionally, the company said account owners or administrators enable the AI features, and participants receive notice through the user interface that AI technology is in use.
Adam Janofsky is the founding editor-in-chief of The Record by Recorded Future. He previously was the cybersecurity and privacy reporter for Protocol, and prior to that covered cybersecurity, AI, and other emerging technology for The Wall Street Journal.