WorldAgents can process audio, video, images, files, text, location-related context, device signals, spatial maps, semantic layers, and simulator events. This processing is used to provide agent responses, recognition, localization, simulation, debugging, and related product features.
We may send content and related metadata to model providers, cloud infrastructure, device services, and other service providers as needed to operate WorldAgents. We use contractual, technical, and operational controls intended to limit processing to the purposes described in this policy and our agreements.
Unless a separate agreement, product setting, or written permission says otherwise, we use non-public customer content to provide and operate WorldAgents, not to train models for general use outside WorldAgents.
You should not submit sensitive personal information, regulated data, or confidential third-party material unless you have permission and the product, plan, and agreement are appropriate for that data.