Zoom has implemented revisions to its terms of service following online criticism over recent updates to its fine print, which had allowed the use of customer data for AI training.
StackDiary, in a report last weekend, shed light on these changes that seemingly granted the company extensive control over customer data to be used for AI training, a move that went unnoticed as it was quietly rolled out in March.
Subsequently, Zoom responded with a blog post, assuring users that it would not engage in the activities mentioned in the updated terms, and in response to the ongoing backlash, they have now revised their terms again.
The updated terms now explicitly state that AI models will not be trained on consumer video, audio, or chats “without customer consent.”
The controversy partly arose from Zoom’s experimental AI tools, such as the IQ Meeting Summary, which utilizes machine learning for summarizations, and the IQ Team Chat Compose, an AI-powered message drafting feature.
Although account owners are required to provide consent before using these tools in a meeting, additional participants were only given two options: either to accept the terms and join the meeting or reject them and leave the meeting.
The most concerning aspect for many was the explicit permission given to Zoom to use customer data for machine learning and AI, including algorithm and model training.
The lack of an opt-out option raised significant concerns about user privacy and consent, as it effectively meant that Zoom could use customer content to train its AI without their explicit agreement.
The terms allowed Zoom to perform a wide range of actions with respect to the customer content, such as redistribution, publishing, importing, accessing, storing, transmitting, reviewing, disclosing, preserving, extracting, modifying, reproducing, sharing, displaying, copying, distributing, translating, transcribing, creating derivative works, and processing customer content.
In an effort to address the mounting criticism, Zoom’s Chief Product Officer, Smita Hashim, emphasized in the company’s latest blog post that account owners and administrators must provide consent before sharing their data for AI training, and the data is used solely to enhance the performance and accuracy of AI services. She reassured users that even if they chose to share their data, it would not be used for training third-party AI models.
Hashim clarified that while Zoom had permission to utilize customer content to offer value-added services based on that content, the customers would still retain ownership and control over their data.
For instance, if a customer had a webinar they wanted Zoom to livestream on YouTube, Zoom might use the video and audio content for the livestream, but the customer would continue to own the underlying content.
To further clarify their position, Zoom introduced a new section to its terms that unequivocally states they will not employ audio, video, or chat customer content to train their AI models without obtaining customer consent.
This addition to the terms ensures that users are fully aware of their rights and maintain control over how their data is utilized.
Hashim emphasized that Zoom’s objective is to empower account owners and administrators with control over the features and decisions related to their data. The company aims to be transparent about these processes and their impact on different customer groups.
In conclusion, Zoom has taken swift action in response to the backlash regarding their updated terms of service, ensuring that customers’ concerns are addressed and their data is handled responsibly.
The company’s commitment to obtaining customer consent before using their data for AI training, and their dedication to maintaining transparency, aims to restore users’ confidence in the platform’s data handling practices.
By incorporating user feedback and making necessary changes, Zoom is striving to strike a balance between providing valuable AI services and safeguarding user privacy and consent.