Zoom’s new terms of service enable the use of customer data to enhance Zoom AI models.
Zoom AI tools can be trained with customer data according to new terms
In an essential move towards advancing its AI capabilities, Zoom is seeking to train its artificial intelligence models with certain user data, as revealed by the newly updated terms of service.
Zoom’s latest amendment to its terms doesn’t only dwell on software licensing, beta services, and compliance; it dives deeper to disclose a significant aspect of Zoom’s AI strategy. Effective from July 27, the updated terms endorse Zoom’s entitlement to use select facets of customer data to refine its AI or machine-learning models.
The specifics include the “service-generated data” that Zoom is now authorized to employ for training its AI, encompassing details like customer information on product usage, telemetry and diagnostic data, and similar content collected by the company. Uniquely, this update doesn’t offer an opt-out option.
While the utilization of such data is not an unusual practice among companies, the fresh terms undoubtedly signal a calculated advancement toward Zoom’s individual AI aspirations.
Coinciding with this update is the escalating public discourse concerning the degree to which AI services, including Zoom’s AI models, should be built on individuals’ data, even when it’s declared to be aggregated or anonymized. In the wide realm of generative AI, tools like OpenAI’s ChatGPT, Google’s Bard, and Microsoft’s Bing, as well as image-creation systems like Midjourney and Stable Diffusion, rely on vast quantities of internet text or images.
This broader conversation lends context to Zoom’s decision, especially at a time when lawsuits have been emerging from authors or artists claiming to recognize their original work in AI-generated content.
“You consent to Zoom’s access, use, collection, creation, modification, distribution, processing, sharing, maintenance, and storage of Service Generated Data for any purpose, to the extent and in the manner permitted under applicable Law, including for the purpose of … machine learning or artificial intelligence (including for the purposes of training and tuning of algorithms and models),” Zoom’s terms reads.
Continuing its commitment to transparency, Zoom has recently specified that customer content such as messages, files, and documents are not included in the categories used for AI training.
Zoom updated terms of service
In a blog post, Zoom stated clearly that “for AI, we do not use audio, video or chat content for training our models without customer consent.” The crux of this clarification lies in the term “without customer consent.”
A recent report from Stack Diary has drawn attention to two particular sections, 10.2 and 10.4, within Zoom’s terms of service that shed light on how the company can handle user data in relation to AI.
Section 10.2 illustrates that customers consent to Zoom’s handling of “Service Generated Data” for virtually “any purpose,” which explicitly includes “machine learning or artificial intelligence (including for the purposes of training and tuning of algorithms and models).”
Section 10.4 further cements Zoom’s rights by stating that customers “agree to grant and hereby grant Zoom a perpetual, worldwide, non-exclusive, royalty-free, sublicensable, and transferable license” to make use of their data for various applications such as “product and service development,” machine learning, and artificial intelligence.
In an added reassurance to its customers, a newly-inserted statement following section 10.4 emphasizes: “Notwithstanding the above, Zoom will not use audio, video or chat Customer Content to train our artificial intelligence models without your consent.”
In the broader context, Zoom is among several businesses actively promoting new AI-powered tools, reflecting a growing trend in the technology industry. These innovative features include a tool designed to assist users in catching up on missed meetings, and another aimed at aiding the composition of messages within its Slack-like Team Chat app. The company’s nuanced handling of customer data for AI purposes demonstrates an alignment with both technological advancement and user privacy.
Earlier in June, Zoom rolled out two innovative generative AI features — a tool that summarizes meetings and another to assist in composing chat messages. These were introduced on a free trial basis for users, offering them the autonomy to choose whether or not to use them.
Zoom launches a local data storage service – but only for the rich
However, Zoom ensures that when a user opts to enable these features, they must sign a consent form. This agreement explicitly permits Zoom to utilize their individual customer content to train its AI models, reflecting a clear policy of user consent and control.
“Your content is used solely to improve the performance and accuracy of these AI services,” Zoom said in the blog post.
“Zoom customers decide whether to enable generative AI features, and separately whether to share customer content with Zoom for product improvement purposes,” a company spokesperson stated.