In response to concerns that its artificial intelligence (AI) models might be trained using customer calls without proper authorization, Zoom has undertaken a revision of its terms of service. The company emphasized in a blog post that audio, video, and chat data would not be utilized for AI purposes unless consent was granted.

Prompted by user observations of modifications to its terms of service in March that raised worries about potential AI training, the video-calling application took action. The motivation behind the changes was to enhance transparency. Subsequently, in June, Zoom introduced novel AI-driven functionalities, including a feature that allows users to summarize meetings without the necessity of recording the entire session. These features were initially made available through a free trial.

Nevertheless, certain experts cautioned against the original wording of the terms of service, suggesting that it could have potentially granted Zoom access to more user data than necessary, including information from customer calls. Prior to the update of the terms of service, Robert Bateman, a specialist in data protection, noted that the language of the terms seemed to provide the service provider with considerable latitude to employ user-generated data for various purposes.

Bateman indicated that while uncertainties existed regarding the possible risks that might arise, the presence of broad contractual provisions such as these should serve as a cause for concern.

On Monday evening, Zoom made a late update to its terms, adding a statement that reads, "Without your consent, Zoom will refrain from utilizing customer content in the form of audio, video, or chat to train our artificial intelligence models."

Privacy Concerns Surrounding AI Applications and Zoom's Approach

AI applications, computer programs or tools capable of performing intelligent tasks traditionally executed by humans, rely on substantial datasets and algorithms for training, enabling them to emulate human-like behavior patterns.

However, the widespread collection of online data for training AI models has ignited apprehensions and led to legal actions due to potential inclusion of personal, sensitive, or copyrighted materials in the datasets.

In response to the escalating enthusiasm for AI technology, Zoom, like numerous other tech companies, has increased its emphasis on AI products this year.

Nonetheless, the Open Rights Group, a digital privacy advocacy organization, has expressed concerns over Zoom's introduction of AI features through a free trial with an "opt-in" approach, deeming these changes disconcerting.

Abby Burke, Policy Manager for Data Protection at the Open Rights Group, noted the ambiguity of Zoom's privacy policy and the lack of clarity regarding customer consent for training AI models, prior to the recent update to Zoom's terms.

A Zoom spokesperson reiterated on Monday that customers retain the choice to enable generative AI features and decide whether to share customer content with Zoom for product enhancement purposes.

Screenshots accompanying the blog post on Monday illustrated warning messages for users participating in meetings using AI tools, presenting the option to either consent to training use or exit the meeting.

Smita Hashim, Zoom's Chief Product Officer, stated that account owners and administrators have the discretion to activate the trial-based features, and those who do so will encounter a transparent consent process for training AI models using customer content.