Zoom has revised its terms of service to state that it will not utilize users’ content without their permission to train its artificial intelligence (AI) tools for the videoconferencing platform.
The company made the changes amid growing concerns about potential privacy invasions and surveillance on Zoom.
Smita Hashim, Zoom’s Chief Product Officer, stated that the updated guidelines make it clear that the company will not use people’s audio, video, or chat data to train their AI models without consent.
Zoom will inform participants when their generative AI services are in use and will provide options for customers to choose whether they want their content used for training AI.
“Your content is used solely to improve the performance and accuracy of these AI services,” wrote Ms. Hashim. “And even if you choose to share your data, it will not be used for training any third-party models.”
Privacy concerns on Zoom extend beyond AI. The platform has faced allegations of enabling espionage by the Chinese government and hackers.
China’s attempts to infiltrate Zoom increased during the COVID-19 pandemic, according to author Bethany Allen-Ebrahimian in the book “Beijing Rules.” She claimed that the country’s security services sought to gain control over the platform and block users around the world.
For instance, China’s Ministry of State Security allegedly tasked Zoom employee Julien Jin with monitoring individuals outside of China. Jin was charged by the Justice Department in 2020 and is wanted by the FBI.
Zoom downplayed concerns about Chinese espionage, stating that the events discussed in the book occurred years ago. The company said it has taken significant steps to address challenges when operating in China.
Fears of unwanted surveillance on Zoom go beyond China. Three UK-based researchers recently discovered a method to spy on users’ keyboard activity through Zoom.
The researchers used machine learning tools to accurately identify 93% of keystrokes from the sounds of typing on a keyboard while using Zoom.
Zoom has not responded to questions about the researchers’ work.