Zoom, the popular video conferencing app, has recently revised its terms of service after facing backlash from users and privacy advocates over its AI policy changes. The company had updated its terms in March 2023, but the controversy erupted this past weekend when a Hacker News post highlighted that the changes seemed to give Zoom the right to use the content of video calls to train its AI algorithms.
What were the changes?
The changes that sparked the outcry were mainly in two sections of Zoom’s terms of service. Section 10.2 stated that Zoom could use diagnostic data, such as telemetry and performance information, for purposes including “machine learning or artificial intelligence (including for the purposes of training and tuning of algorithms and models).” Section 10.4 stated that Zoom could use customer-generated content, such as audio, video, and chat, for a list of uses that included “machine learning, artificial intelligence, training, testing.”
Many users and experts interpreted these clauses as giving Zoom the permission to analyze and learn from the content of their video calls without their consent. They raised concerns about the potential privacy implications and ethical issues of using personal and sensitive data for AI purposes.
How did Zoom respond?
Zoom published a blog post on Monday, clarifying its intentions and revising its terms of service. The company said that it had rewritten its terms to be more transparent about how it uses data to provide and improve its services, not to claim new rights to user data. It also said that it does not use audio, video, or chat content to train its AI models without customer consent, and added a new paragraph in bold to section 10.4 to reflect that.
Zoom’s CEO Eric Yuan also admitted in a LinkedIn post on Tuesday that the March terms of service changes were a mistake and that the company had a process failure internally that it would fix. He reiterated that Zoom would seek user consent for any AI training and that it would not share user data with any third parties for AI purposes.
Why does Zoom use AI?
Zoom already uses AI to power some of its features, such as meeting transcription, live captioning, virtual backgrounds, and gesture recognition. It also has some experimental features that use AI, such as meeting summary and follow-up suggestions. The company has outlined some future AI features that it wants to offer, such as using prompts to help start a whiteboard session and using AI to organize the ideas generated by meeting participants.
Zoom says that its AI-powered features are off by default and that users can choose whether to enable them or not. It also says that account owners and administrators can control whether to allow these features for their accounts. They can also opt out of providing data to Zoom for model training and can change the data sharing setting at any time.
What are the challenges and opportunities of using AI for video conferencing?
AI can offer many benefits for video conferencing, such as enhancing the quality and accessibility of meetings, providing insights and feedback, and facilitating collaboration and productivity. However, AI also poses many challenges and risks, such as ensuring the privacy and security of user data, respecting the consent and preferences of users, maintaining the accuracy and fairness of algorithms, and addressing the ethical and social implications of using AI for communication.
As video conferencing becomes more prevalent and essential in the post-pandemic world, Zoom and other platforms will have to balance the opportunities and challenges of using AI for their services. They will also have to be transparent and accountable for their data practices and policies, and engage with their users and stakeholders to build trust and confidence.