One additional concern with the current workaround is privacy. As a user, I want to tap the mic on a third-party keyboard, dictate, stop, and do it again later without switching apps. Because keyboards can’t access the microphone directly, apps can instead keep a microphone session open in their containing app. That gives me the seamless experience I want, but means the microphone can remain available in the background between dictations. My choice is effectively convenience or OS-enforced microphone privacy. This seems backwards. If iOS allowed user-initiated microphone access from a keyboard extension, I could explicitly grant permission and iOS could enforce the boundary: tap mic → microphone on → dictate → stop → microphone off. That would give users both the better UX and much clearer control over when a third-party keyboard can hear them. I’ve filed an enhancement request for this: FB24570791.