During live stream, the model can only be initiated with text or voice

Using the Live stream feature, I was sharing my screen and asking Gemini to notify me when it notices certain behavior - e.g. nudge me to focus when it notices I’m going to Youtube.However, it seems Gemini was not able to proactively speaks back, unless I ask it explicitly via voice or text.

Hi @hugies,

Yes, you’re right. Gemini won’t speak up on its own during a live stream. It only reacts when you actually say or type something. It’s designed that way for safety.