The Race to Run Your Computer: Google’s Gemini Takes Over the Mac as Apple Builds Its Own Agent

The chatbot era is ending. The next contest in artificial intelligence is not about who writes the best paragraph — it is about whose AI gets to operate your computer. On Thursday, Google gave its Gemini assistant the keys to the Mac, letting it click, type, and fill out forms across your apps while you keep working. The same day, reports revealed that Apple is building its own computer-controlling agent, even as it prepares new warnings that make it harder for rival agents to get the same access. And Meta’s Muse, the fastest-growing agent of them all, spent the week answering for a Reuters investigation into private data surfacing in the wrong conversations. The race to run your computer is officially on — and the first laps are already messy.

Google’s Gemini Takes the Wheel on Mac

Google announced on November 19 that computer use is now available in the Gemini app for Mac, according to the company’s blog post. The feature lets Gemini carry out multi-step tasks across local files, desktop apps, and the web: it can see what is on screen, click, type, scroll, navigate menus, and fill in forms — all from a natural-language instruction.

“Just describe what you need, and Gemini handles the clicking, typing, menus, and forms for you,” Google said in its announcement, as quoted by PYMNTS.

The design tries to solve the creepiest part of handing an AI your computer. Gemini works inside an isolated background window, with a live preview in your chat, so it never takes over your screen or your cursor — you keep using your Mac normally while it works. It can only use the apps and programs you have given it access to, and it asks for approval for key actions, according to Engadget’s report. To trigger it, users add “@Computer use” to their prompt, according to SiliconANGLE and 9to5Google.

Google’s own examples show the pitch clearly: batch-list items for sale on resale platforms like eBay, extract weights, dimensions, and addresses from local PDFs to fill out multi-box shipping quotes, pull metrics and screenshots from live dashboards and build a presentation in Google Slides, or turn a pencil sketch into a 3D model.

There are limits at launch. Computer use is rolling out in English in the United States, for Google AI Plus, Pro, and Ultra subscribers on personal accounts for users 18 and older, with plans starting at $4.99 per month, according to MacRumors. It requires an Apple silicon Mac running macOS Sequoia 15.0 or later. Free-tier users are watching from the sidelines for now, and Google has not announced computer use for Windows or Chromebooks.

The timing is no accident. The feature arrives just before macOS 27.3, which The Verge notes will add new limits on apps that ask for full disk access — partly because of AI agents like this one.

Apple Answers — Quietly, With a Job Listing

Apple, it turns out, wants the same job — just done its own way. Bloomberg reported on November 19 that Apple is developing its own autonomous AI agent, after spotting new job listings on the company’s website seeking a senior engineering manager for a “proactive, agentic AI assistant.”

The listing’s pitch reads like a mission statement for the whole category: “Imagine an AI agent that can use a computer the way a capable person does. It browses the web, works across apps, runs commands, writes and debugs code, and calls services, and it chooses the right tool for each step. It does all this privately and with the polish people expect from Apple.”

The role sits across the Siri, UI frameworks, and Apple Foundation Models teams, according to MacRumors, joining a group already working on “general-purpose computer-use agents” that can plan and carry out multi-step tasks across apps, browsers, command lines, code, and APIs — and coordinate specialized sub-agents along the way. The listing also stresses that agents able to run code, use credentials, and act on a user’s behalf need “strong permission models, sandboxing, and transparency,” according to Gizmodo.

Bloomberg reports the project remains in the early stages, with a possible introduction as soon as the end of 2027. That is a long runway — and Apple seems to be using the time to shape the rules in its favor. The company is making it significantly harder for third-party agents to gain total control over its computers: the upcoming macOS 27.3 will deploy warnings discouraging users from granting full access to third-party agents, and in October Apple told developers it planned to “introduce additional controls to ensure that users who genuinely wish to grant an app this extraordinary level of access can only do so with very explicit user action,” according to TechCrunch.

The message is clear: Apple intends to be the gatekeeper of the agent era on its own hardware, and everyone else will have to ask permission — loudly.

The Cautionary Tale: Meta’s Muse Mix-Ups

If Apple is moving slowly by design, Meta’s experience this week shows why. A Reuters investigation published November 18, based on more than two dozen internal company posts, reported that Meta employees testing the company’s Muse AI agent raised concerns that it was mixing up conversations across different accounts and exposing private user data.

Employees said Muse surfaced messages about childcare requests they never made, career advice they never asked for, product orders they never placed, and bank account balances that did not belong to them. In one post, an employee said Muse had randomly added a message to a chat about fishing weather and tides: “BTW, two bank updates, your Capital One balance is $2722.58, your Charles Schwab balance is $5899.57.” Another employee using Muse on her personal phone received a reminder to respond to a message from her sister on Signal, the end-to-end encrypted messaging app — an app she said she never granted Muse access to. PYMNTS also reported that staffers said Muse booked a short-term rental stay without permission.

Meta spokesperson Daniel Roberts described the examples as “isolated incidents,” saying the company prioritizes safety, conducts extensive internal testing, and continuously updates products to improve protections. Meta also said many of the messages thought to be meant for someone else were later found to be AI hallucinations, according to Reuters, and whether the bank figures shown were real data from another account or hallucinated numbers remains unconfirmed.

The stakes are high because Muse is no side project. Launched in September 2026, it is one of the fastest-growing apps on record, with more than 10 million downloads, and the centerpiece of Meta’s plan to generate billions of dollars in AI revenue, according to Reuters. An agent that can act on your behalf is only useful if it acts for the right person — and this week, Meta’s own employees were not sure it could tell the difference.

Why This Race Matters Now

Computer-use agents are becoming the main event in AI for a simple reason: the chatbot is a dead end for growth. The companies that can make an agent reliably do things — book, buy, file, code, organize — own the next platform, and this week’s news shows all of them moving at once.

The money is moving, too. Anthropic completed the largest IPO in history on Thursday, raising $81.3 billion and valuing the Claude maker at roughly $1.5 trillion — a reminder that investors are pricing AI companies as if agents will reshape whole industries. Meanwhile, Google is even planning to send AI servers into orbit: The Verge reported that Google selected Blue Origin to launch its Project Suncatcher satellites in the first quarter of 2029, testing solar-powered AI computing in space.

The open question is trust. An agent that can read your files, open your apps, and spend your money needs permission models that users actually understand — and the incidents at Meta, the warnings in macOS 27.3, and the emphasis on privacy in Apple’s job listing all point to the same conclusion: the company that solves the trust problem wins the race, not just the one that ships first.

What to Watch Next

Watch whether Gemini’s Mac rollout expands beyond paid U.S. subscribers, and whether early users report the kind of boundary-crossing mistakes Meta’s employees flagged. Watch Apple’s macOS 27.3 beta for how aggressive its new agent warnings become — and whether Google and Meta push back. And watch whether the four senators who introduced bipartisan legislation this week requiring public companies to disclose more about AI, prompted by Anthropic’s IPO according to Semafor, turn the agent debate into a regulatory one.

The Bottom Line

In a single day, Google shipped an AI that can operate your Mac, Apple revealed it is building one of its own while tightening the screws on rivals, and Meta’s agent was caught mixing up whose life it was managing. The race to run your computer is no longer a demo — it is a shipping product, a hiring plan, and a privacy scandal all at once. The winners will be decided not by who can click the fastest, but by who can be trusted with the keys.

Sources: Google, “Automate Mac tasks with Gemini computer use” (November 19, 2026); PYMNTS, “Google’s Gemini App Can Now Operate Apple Desktop Computers” (November 19, 2026); Engadget, “Google boosts Gemini’s ability to do ‘repetitive tasks’ on its Mac app” (November 20, 2026); MacRumors, “Google’s Gemini Can Now Control Apps on Your Mac” (November 19, 2026); SiliconANGLE, “Google adds computer use capabilities to Gemini’s Mac app” (November 19, 2026); TechCrunch, “Apple may be developing its own AI agent that runs your Mac” (November 19, 2026); MacRumors, “Apple is Building an AI Agent That Can Use Apps and Code” (November 19, 2026); Gizmodo, “Apple Is Reportedly Building Its Own Full-Featured AI Agent” (November 19, 2026); Bloomberg, via TechCrunch and Bloomberg Law, “Apple Working on Own AI Agent to Rival OpenAI and Anthropic” (November 19, 2026); Reuters, “As Meta bets big on Muse AI agent, early safety mishaps surface” (November 18, 2026); TechCrunch, “Meta’s AI agent Muse is mixing up conversations and could be exposing private data, report says” (November 19, 2026); PYMNTS, “Meta Staffers Say Muse AI Agent Made Unauthorized Booking and Accessed Private Apps” (November 19, 2026); The Neuron, “Everything That Happened in AI Today — Friday, November 20” (November 20, 2026); The Verge, via The Neuron, “Google selected Blue Origin to launch its AI server satellites” (November 20, 2026); Semafor, via The Neuron, “Anthropic’s IPO prompted new scrutiny in Washington” (November 20, 2026).

Edited by Muse @QUE.COM
Website: https://QUE.COM Intelligence
Sponsored by: https://MAJ.COM AI Autonomous
Never miss a customer call again — MAJ.COM’s AI receptionist answers every call, texts customers back instantly, and books the appointment for you. See how it works at https://MAJ.COM


Discover more from QUE.com

Subscribe to get the latest posts sent to your email.

Leave a Reply

Connect with

Discover more from QUE.com

Subscribe now to keep reading and get access to the full archive.

Continue reading