Google Gemini Gains Mac Control, Prompting Apple Security Overhaul

3 min read
Source: MacRumors
Google Gemini Gains Mac Control, Prompting Apple Security Overhaul
Photo: MacRumors
TL;DR

Google has launched 'computer use' for its Gemini app on macOS, allowing the AI to autonomously navigate apps, fill forms, and manage files in the background. The feature requires Apple Silicon Macs and a paid subscription, while Apple simultaneously introduced stricter permission controls in macOS 27.3 to mitigate risks associated with AI agents.

Key points

  • Gemini can now perform multi-step tasks like batch editing photos or filling shipping forms across local apps and the web.
  • The feature operates in an isolated background window with a live preview, allowing users to continue working in the foreground.
  • Availability is limited to the U.S. in English for Google AI Plus, Pro, and Ultra subscribers with Apple Silicon Macs running macOS Sequoia 15 or later.
  • Apple introduced new permission controls in macOS 27.3 beta, including indicators for app control and options to restrict access to specific folders.
  • Users can grant 'Always allow' permissions, though Google states sensitive actions still require additional confirmation.

Background

This development follows Google's recent expansion of AI capabilities, including the rollout of Gemini on Apple CarPlay and the integration of AI-generated apps in the Play Store. It also aligns with Apple's ongoing efforts to secure its ecosystem, following warnings about the risks of Full Disk Access for AI agents.

How outlets are covering it

MacRumors and The Verge highlight the convenience of Gemini's new automation capabilities, noting its ability to handle repetitive tasks like creating presentations or listing items for sale. In contrast, AppleInsider emphasizes significant security risks, warning that granting broad permissions could expose users to prompt injection attacks and unauthorized access to sensitive accounts. While Google frames the feature as a productivity tool with built-in safeguards, AppleInsider argues that the potential for irreversible changes and data exposure makes broad AI control on a primary Mac inadvisable. All sources agree that Apple's recent macOS updates aim to provide users with greater transparency and control over app permissions.

Why it matters

The introduction of autonomous AI agents on macOS marks a shift from conversational assistants to active digital workers. This trend intensifies the competition between Google and Apple, forcing the latter to implement stricter security measures to protect user data. The feature's availability to paid subscribers also signals a new monetization strategy for advanced AI capabilities, potentially influencing how software companies price automation tools in the future.

What to watch

Users should monitor the rollout of macOS 27.3 to understand how new permission controls affect existing AI applications. Google may expand computer use to other regions and languages, while Apple could introduce further restrictions on AI agents in future updates. Users are advised to review their app permissions and consider limiting AI access to non-critical tasks to mitigate security risks.

Share this article

Want the full story? Read the original reporting

Read on MacRumors