Videoconferencing company Zoom has clarified a policy that had some users nervous. It will not be using its calls to train AI.

A recent terms of service update implied customer video calls could be used to train AI models. Those terms said that “service generated data” and “customer content” could be used “for the purpose of product and service development,” such as “machine learning or artificial intelligence (including for the purposes of training or tuning of algorithms and models.”

There was a bit of real-world uproar at that.

Zoom Chief Product Officer Smita Hashim walked it back in a blog post that “[Zoom does] not use audio, video, or chat content for training our models without customer consent.” She indicated that Zoom customers own data like meeting recordings and invitations, and that “service generated data” referred to telemetry and diagnostic data and not the actual content of customers’ calls.

Zoom has now updated both the terms of service and Hashim’s blog post, and each now contains the same statement in bolded text:

“Zoom does not use any of your audio, video, chat, screen sharing, attachments or other communications-like Customer Content (such as poll results, whiteboard and reactions) to train Zoom or third-party artificial intelligence models.”