(The Defender) When Zoom updated its terms of service (TOS) in late July, the tech company sparked criticism and outrage among cybersecurity experts and users concerned about Zoom using customers’ data — without their consent — to train artificial intelligence (AI).
In an attempt to allay fears about privacy infringement, Zoom revised its TOS — on Aug. 7 and again on Aug. 11, publishing a companion blog post to clarify its approach to AI.
Zoom confirmed three times in its blog post, “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.”
According to Zoom, the AI features are turned off by default. However, if a meeting administrator — typically the host — turns them on, the only way for participants to opt out is to leave the meeting.
Referencing a January ruling by the Court of Justice of the European Union against Meta’s forced consent practices, TechCrunch reported, “It goes without saying that telling your users the equivalent of ‘hey, you’re free to leave’ does not sum to a free choice over what you’re doing with their data.”
Data are ‘the new gold’
Zoom states that it retains all rights to service-generated data, which means the company can modify, distribute, process, share, maintain and store such data.
Attorney W. Scott McCollough, whose experience includes writing TOS and acceptable use policies for the tech/telecommunication industry, told The Defender:
“Let us not be fooled and this should be no surprise. Zoom is freeware, although it has [a] subscription, too.
“When the ‘service’ is free the user is the product. There is always a route to user data appropriation and monetization no matter what the terms imply or what boxes the user clicks.”
Although there are reams of federal and state privacy laws, the exception to all of them is consent, attorney Greg Glaser told The Defender:
“There’s a huge value in data. Data is the new gold. Zoom is trying to get around the invasion of privacy laws by saying you consented to Zoom using and training AI on your own video [call].”
The ultimate goal? To use advanced AI to create experiences that can be sold, said Glaser.
‘They have big plans for our data’
According to Search Logistics, Zoom logged 10 million daily meeting participants in December 2019. Since then, Zoom has grown by over 2,900% and currently averages 300 million daily active users.
According to Glaser, Zoom — along with other tech behemoths like Meta — is trying to put into terms early that they have legitimate access to consumer information. Then they would have carte blanche to create and use “emulates,” a computer version of an actual person.
Glaser added:
“There’s this huge demand for real people. They [Zoom and other tech companies] are trying to expand terms of use so they can share your information, so they can play with it. It’s the first step in the ultimate direction they want to go, which is emulates.
“It is no accident that organizations like Meta, Zoom and Google utilize AI on our private videos. They have big plans for our data, and we’re not entirely necessary in their weird view. According to Robin Hanson in [his] TedTalk about emulates [or “ems” — digital copies of human minds] … ‘Your job is to retire and die.’
“What we’re experiencing is a great showdown between the forces of Western civilization and neo-dystopia.”
And don’t forget Zoom’s ownership structure, said McCollough.
“They are public, although management insiders own almost 60%. Among them is Lt. Gen. (Ret.) H.R. McMaster, former national security advisor to Trump famously tweet-fired for not doing enough to end the forever war in Afghanistan.
“There are also some familiar faces in the Class A lineup: Vanguard and T. Rowe Price. Both big in Pharma. Vanguard is big in telecom, too. The Class B shares appear to be held by others large into surveillance capitalism.”
What are these AI features?
In March, Zoom announced its generative AI offerings, which use neural networks to identify the patterns and structures within existing data to generate new and original content.
