Gemini 3.5 Flash Adds Native Computer Use Capabilities for Enterprise Automation

Google has integrated computer use functionality directly into Gemini 3.5 Flash, enabling developers to build intelligent agents capable of cross-platform automation. Previously available only as a standalone model variant, this feature now comes built into the main Flash model, expanding its capabilities for complex agentic tasks. The enhancement targets enterprise use cases, particularly continuous software testing and knowledge work automation across professional applications. Developers and organizations can access this capability through the Gemini API and Gemini Enterprise Agent Platform. Security has been a design priority. Google applied targeted adversarial training to protect against prompt injection attacks and offers optional enterprise safeguard systems. The company recommends combining these security features with additional measures such as secure sandboxing and human oversight for deployment in production environments.