A Mac computer screen showing the Gemini AI interface performing automated tasks in a separate window.
Crypto

Google Gemini Gains Mac App Control With New Computer-Use Feature

Illustration: NexVani Media

By: NexVani DeskDate: 20 November 2026Views: 0

Google has launched a new computer-use feature for Gemini on Mac, allowing the AI to perform tasks across apps, though the update introduces significant security and privacy considerations.

This is news, not investment advice. NexVani Media is not a SEBI-registered investment adviser and does not recommend buying or selling any security. Do your own research or consult a registered adviser before investing.

Google has expanded the capabilities of its AI assistant, moving beyond simple text responses to active desktop management. Users searching for Google Gemini can now utilize a new computer-use feature on Mac that allows the AI to navigate menus, open applications, fill out forms, and transfer data between local files and web interfaces. This update represents a shift toward agentic AI that performs manual tasks like typing and clicking on behalf of the user.

Understanding Google Gemini on Mac

The feature operates within an isolated background window, which allows the AI to execute multi-step workflows without seizing control of the user's primary cursor or display. According to Google, this design enables owners to continue working on their main desktop while the agent handles secondary processes, such as extracting shipping details from PDFs, organizing photo batches, or creating presentations from dashboard data. To activate the tool, users must enable computer use in the Mac app and initiate tasks using the @Computer use command.

To function, the system requires users to grant macOS Accessibility and Screen & System Audio Recording permissions. While Google provides mechanisms to create lists of allowed or blocked applications and allows users to monitor the agentโ€™s progress via a live preview, the integration carries inherent risks. Googleโ€™s documentation explicitly warns that the agent can access active sessions, view authentication codes on the screen, and perform irreversible actions within software.

Google's new Gemini Mac feature automates desktop tasks but warns users about potential prompt injection security risks.

Furthermore, the system is susceptible to prompt injection attacks. Malicious instructions embedded in websites or documents could potentially trick the AI into executing unauthorized commands across different applications. Google suggests that while some sensitive actions require manual confirmation, users must remain vigilant regarding the permissions granted to the AI. The feature is currently available in English for adult users in the United States with personal paid accounts, provided they are using an Apple Silicon Mac with macOS Sequoia 15.0 or later.

Why it matters

This development marks a significant step in the evolution of AI agents that interact directly with local operating systems rather than just cloud-based data. By automating complex, multi-app workflows, Google aims to increase productivity for Mac users who manage heavy administrative tasks.

However, the reliance on deep system permissions and the potential for security vulnerabilities like prompt injection highlight the trade-off between AI-driven convenience and personal data privacy. Users must weigh the efficiency of automated desktop control against the risk of granting an AI agent access to sensitive accounts and local files.

Add NexVani Media as a preferred source on Google

Sources

ND

NexVani Desk

NexVani Media brings you the most important world news as it happens, with every story linked to its sources.