Asking as an individual developer, before building further. Relevant guidelines: 2.5.14, 5.1.1, 5.1.2.
Remote support apps are established on the App Store: the user starts a broadcast with RPSystemBroadcastPickerView and a Broadcast Upload Extension, and a human agent views their screen to help them.
My design is identical except the frames go to an automated analysis service instead of a human. The results return only to the same user who started the session, to help with their own task on their own device.
The design: sessions are started and ended by the user through the system picker, with the standard broadcast indicator visible throughout. Frames are transmitted transiently, never stored, and never seen by any person. There is no remote control or synthetic input of any kind. Captured data is used only for the in session assistance, with explicit consent and privacy policy disclosure per 5.1.2(i).
Is swapping the human viewer for an automated analyzer acceptable under the guidelines? And does automated processing of screen content require any disclosures or consent beyond what human viewing requires?
Thanks for any guidance.