Zoom, the popular video conferencing platform, is facing scrutiny over its terms of use, which include provisions related to data usage for artificial intelligence (AI) training. While sections on software licensing, beta services, and compliance are placed at the beginning of the terms, a closer look reveals a clause that allows the company to utilize customer data for AI training without an explicit option to opt out.

The language used in Zoom’s terms and conditions is both intricate and deliberate. Users agree to a broad range of data-related activities, including accessing, using, modifying, distributing, processing, and storing data generated by the service, all within the limits of applicable laws. This includes data utilization for the purpose of AI training and customization of algorithms and models.

Experts have noted that the document specifically references “service-generated data,” which covers telemetry data, product usage data, diagnostic data, etc., but not necessarily all user-generated content such as audio, video, or chat content. However, an important addition in the document’s appendix states that Zoom “does not use audio, video, or chat content to train AI models without customer consent.” The critical phrase here is “without the consent of the client,” implying that by using the program, users implicitly consent to these terms, often without fully reading them.

This revelation comes at a time when concerns about AI’s use of data for learning are mounting. Various AI models, including chat bots and image generators, learn from snippets of text and images taken from the internet, raising questions about intellectual property rights and content ownership.

In recent months, there has been an increasing number of lawsuits from authors and artists who claim that their work is being mirrored in AI-generated results. The introduction of new generative AI features by Zoom has further heightened these concerns. The company launched tools for creating meeting summaries and chat messages using AI, but users enabling these features are prompted to sign a consent form during the conversation. This form allows Zoom to train its AI models with individual customer content. The user has the option to agree by clicking “Got it!” or to leave the meeting. This dynamic essentially puts participants in a position to consent or terminate the conversation.

Aside from potential reputational risks, Zoom might face legal challenges, especially within the European Union (EU), where data protection laws hold significant weight. The General Data Protection Regulation (GDPR) and the updated Electronic Privacy Directive (ePrivacy) could come into play. The latter now includes Zoom-like services and prohibits intercepting or surveilling communications and data without user consent.

Zoom’s press office provided a statement to the press that echoed the intricate wording of the terms of use, leaving the situation open to further interpretation and analysis. As AI technologies continue to evolve, the intersection of data usage, privacy, and consent is likely to remain a contentious topic, prompting closer examination and regulation.