Apple may be developing its own AI agent that runs your Mac

Glass Epoch DeskOriginal digestSource: TechCrunch AI

AI-assisted digest, checked against the original source before publishing. How we work

A job listing suggests Apple is building a computer-use agent, even as it tightens controls on third-party rivals.
Apple may be developing its own AI agent that runs your Mac
Photo: Laurenz Heymann · free stock license

Apple appears to be building an AI agent that can operate a computer on a user's behalf — opening apps, browsing websites, running commands, and writing and debugging code. The signal came from a job listing the iPhone maker recently posted for a senior engineering manager to lead work on an "AI agent that can use a computer," as first reported by Bloomberg.

The posting spells out the ambition in detail. It asks candidates to imagine an agent that works "the way a capable person does" — selecting the right tool for each step, moving fluidly between tools, and carrying out multi-step tasks across apps, browsers, command lines, code, and APIs. It also promises the work happens privately, "with the polish people expect from Apple." TechCrunch said it contacted Apple for comment but did not hear back.

The move would place Apple alongside a fast-growing field. Anthropic and OpenAI have both launched computer-use services, and Meta recently introduced its Muse agent line. The listing arrives as Apple seeks to expand its AI offerings under hardware chief John Ternus, who now leads the company's efforts.

At the same time, Apple is making it harder for outside agents to take the same kind of control of its machines. Bloomberg reports that an upcoming macOS 27.3 release will show warnings discouraging users from granting third-party agents full access to their computers. 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." The pattern suggests a platform strategy: Apple's own agent gets privileged footing while rivals face added friction.

Timing, though, is far from settled. Bloomberg writes that Apple has eyed launching the service as soon as the end of next year, but that internal progress remains in the early stages — meaning the job listing reflects direction more than a shipped product. The key open questions: whether the additional macOS controls apply equally to Apple's own agent or only to third-party apps, how developers will be affected by the new permission flow, and whether Apple's privacy positioning becomes a differentiator against OpenAI, Anthropic, and Meta in a category where users are being asked to hand over deep system access.

Key facts

  • • Apple posted a job listing for a senior engineering manager to build a computer-use AI agent.
  • • The agent would browse the web, run commands, write and debug code, and call services.
  • • Bloomberg reports Apple targets a launch as soon as the end of next year, but work is early.
  • • macOS 27.3 will warn users against granting third-party agents full computer access.
  • • Anthropic, OpenAI, and Meta (with Muse) already offer computer-use agents.

Why it matters

Apple entering computer-use agents legitimizes the category for its consumer base of hundreds of millions of Mac and iPhone users, raising competitive pressure on OpenAI, Anthropic, and Meta. But the simultaneous tightening of third-party permissions in macOS signals a gatekeeper play: builders of rival agents may face an explicit-consent wall on Apple hardware before Apple's own agent ships, a dy

Spotted an error? Send a correction.

Get the day's AI news in one email

The Daily Epoch · 60-second read · unsubscribe anytime

Related

Agents
TechCrunch AI2 min read

Superhuman pushes its Go assistant everywhere

Superhuman, formerly Grammarly, is making its Go assistant the default AI layer across Docs and Mail, adding cross-app context, 20 third-party agents, and proactive drafting.