Zoom reverses policy that allowed it to train AI on customer data
Five months after publishing the terms, the company updated them only after an online outcry.
Zoom has made changes to its terms of service after online blowback over recent updates to the company’s fine print allowing AI training on customer data. A report from StackDiary over the weekend highlighted how the changes, which rolled out in March without fanfare, appeared to grant the company sweeping control over customer data for AI training purposes. In response, Zoom published a blog post today claiming it wouldn’t do what its terms said it could do; the company then updated its terms in response to the continued blowback. It now says it doesn’t train AI models on consumer video, audio or chats “without customer consent.”
At least part of the issue stemmed from Zoom’s experimental AI tools, including IQ Meeting Summary (ML-powered summarizations) and IQ Team Chat Compose (AI-powered message drafting). Although account owners have to provide consent before starting a meeting using these tools, additional participants are only presented with two options: accept the terms and join the meeting, or reject them and leave the meeting.
“What raises alarm is the explicit mention of the company’s right to use this data for machine learning and artificial intelligence, including training and tuning of algorithms and models,” Alex Ivanovs wrote for Stack Diary. “This effectively allows Zoom to train its AI on customer content without providing an opt-out option, a decision that is likely to spark significant debate about user privacy and consent.”
Ivanovs highlighted how the terms give it the right to “redistribute, publish, import, access, use, store, transmit, review, disclose, preserve, extract, modify, reproduce, share, use, display, copy, distribute, translate, transcribe, create derivative works, and process Customer Content and to perform all acts with respect to the Customer Content.”
In a company blog post published Monday, Zoom’s Chief Product Officer Smita Hashim stressed that account owners and administrators indeed have to provide consent before choosing to share their data for AI training, insisting it’s “used solely to improve the performance and accuracy of these AI services.” Hashim added that “even if you chose to share your data, it will not be used for training of any third-party models.” She continued: “We have permission to use this customer content to provide value-added services based on this content, but our customers continue to own and control their content. For example, a customer may have a webinar that they ask us to livestream on YouTube. Even if we use the customer video and audio content to livestream, they own the underlying content.”
“We will not use customer content, including education records or protected health information, to train our artificial intelligence models without your consent,” the blog post reads. A new section added to Zoom’s terms today makes it clearer: “Notwithstanding the above, Zoom will not use audio, video or chat Customer Content to train our artificial intelligence models without your consent.”
“Our goal is to enable Zoom account owners and administrators to have control over these features and decisions, and we’re here to shed light on how we do that and how that affects certain customer groups,” Hashim wrote.