Zoom, the video conferencing company that emerged in 2020 during the Covid epidemic, is receiving harsh criticism for its revised terms and practices. Following two changes to its AI approach, the corporation has announced that it will develop AI-powered tools using client information without user approval. Zoom claims complete authority over data collected during Zoom calls under the amended AI policy terms. Furthermore, they can reuse and share this data in any way they wish while adhering to the law to develop their artificial intelligence/machine learning frameworks. By doing so, they have failed to give users the choice to opt out.
As part of our commitment to transparency and user control, we are providing clarity on our approach to two essential aspects of our services: Zoom’s AI features and customer content sharing for product improvement purposes. Our goal is to enable Zoom account owners and…
— Zoom (@Zoom) August 7, 2023
Updated TOS with AI policy
Zoom revised its service terms in July. These policy revisions follow a public uproar over ethical implications of developing AI models with consumer data. According to the TOS, Zoom may collect and use “service-generated data” on product usage, telemetry, and inspections to train AI models. Customers agree to allow Zoom to access, utilize, gather, create, change, distribute, process, manage, and keep service-related data for any reason, including ML or AI training and algorithm tuning, without attribution. The TOS statement, on the other hand, fails to explain or clarify what Zoom means by service-generated data.
Furthermore, according to revised policies, Zoom has obtained a long-term and global license that allows them to freely share, issue, access, utilize, stock, pass on, evaluate disclose, protect, extract, change, replicate, share, exhibit, distribute, translate, record, generate fresh versions, and manage customer content. A non-exclusive and royalty-free license allows sublicenses and transfers.
People express concerns about openness and privacy laws. Many people are concerned that Zoom’s updated restrictions may have significant ramifications in the telehealth and educational industries. These industries are governed by privacy laws. They also believe that it is critical to address AI model training issues using data gathered from recorded sessions.
Meanwhile, it denies using user information for AI training without first obtaining authorization. According to a blog post by Zoom, clause 10.2 is merely to enhance openness around data usage, to improve customer experience. This would entail studying usage trends such as peak hours in specific time zones to optimize the data center. This would improve video quality. Although Zoom has stated that it will not use audio, video, or chat content for training its models in the healthcare and educational sectors without client authorization, its terms of service appear inconsistent.
Smita Hashim, Chief Product Officer for Zoom writes,
An example of a machine learning service for which we need license and usage rights is our automated scanning of webinar invites/reminders to make sure that we aren’t unwittingly being used to spam or defraud participants. The customer owns the underlying webinar invite, and we are licensed to provide the service on top of that content. For AI, we do not use audio, video, or chat content for training our models without customer consent.
Zoom’s AI-powered assistant Zoom IQ contradicts TOS
A few months ago, Zoom introduced Zoom IQ, an AI-based assistant. This function, which summarizes conversation threads and produces automated answers to chat queries, is optional. However, Zoom IQ is automatically enabled. Users who do not change their settings give businesses permission to collect their data.
The other participants in the call receive a notice stating “Meeting Summary has been enabled” when a Zoom meeting starts with Zoom IQ enabled. Inputs and AI-generated content can be accessed from all users by the account holder. The company ties this to model training and product improvement. The participants have two options: accept the meeting or quit it. It is impossible to avoid sharing private information. In other words, someone else can give Zoom permission to use your data to create AI on your behalf.
A look at the future
Zoom, whose popularity surged during the Covid-19 outbreak, has received criticism for its privacy practices in the past. The business was hit with a $85 million lawsuit the year before due to a security flaw that let hackers access unidentified virtual meeting rooms. Making adjustments to its policies to remain relevant is unavoidable given the constantly changing AI landscape. In addition to OpenAI’s ChatGPT and Google’s BARD chatbot, other AI platforms have raised similar concerns. The results of Zoom’s recent changes to its terms and rules are uncertain. Will it further damage its well-established reputation or turn out well?