Computer use becomes a built-in Gemini 3.5 Flash tool

Google added built-in computer use to Gemini 3.5 Flash on 24 June, moving the capability beyond its separate 2.5 computer-use model. Developers could access it through the Gemini API and Gemini Enterprise Agent Platform to build agents working across browser, mobile and desktop interfaces.

The release also described optional enterprise safeguards: explicit confirmation before sensitive or irreversible actions and automatic stopping when an indirect prompt injection is detected. Google still recommended sandboxing, strict access controls and human review. That combination is more informative than a claim that an agent can simply operate any application safely. A screen-action capability expands the tasks developers can attempt; the surrounding permissions and verification determine which actions an application should actually allow.