Apple's job posting reveals an autonomous AI agent program, aiming to launch as early as the end of next year
BIBIBI
AT A GLANCE
According to Bloomberg, Apple is developing autonomous AI agents that can perform tasks across apps, with a target launch as early as the end of next year.
Article
Apple is developing its own AI agent: capable of controlling devices, browsing the web, and writing code
PANews November 20 reports that, according to Bloomberg, Apple is developing an autonomous AI agent that can write code, browse the web, control apps, and operate devices on behalf of users. Job postings published on Apple's website this week show the company is hiring a Senior Engineering Manager for a "proactive, agent-based AI assistant," with the goal of launching the service as early as the end of next year. The report notes that Apple has assembled a team to build the foundation for this AI agent and plans to enable it to plan and execute multi-step tasks across apps, browsers, command lines, and APIs. Privacy is one of the core elements of this system; Apple states it will incorporate robust permission models and sandboxing mechanisms. Meanwhile, in the upcoming macOS 27.3 beta, Apple has added a new warning screen requiring third-party AI tools to obtain user consent before controlling Macs, a move that has prompted some AI companies to...
PANews reports that, according to Bloomberg, Apple is developing its own AI agent that can write code, browse the web, control apps, and operate devices on behalf of users.
02
The report noted that Apple posted a job listing on its website this week, seeking a Senior Engineering Manager for an “proactive, agent-based AI assistant.”
03
The report states that Apple has formed a team to build the foundation of this AI agent and plans to enable it to perform multi-step tasks across apps, browsers, the command line, and APIs.
04
The report lists privacy as one of the system’s core elements and states that Apple said it will build in a permissions model and sandboxing mechanisms.
05
According to the report, Apple has added a warning screen in the upcoming macOS 27.3 beta, requiring third-party AI tools to obtain user permission before controlling a Mac.
06
The material states that the service is targeted for launch as early as the end of next year.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
The report states that Apple is developing an AI assistant capable of operating devices and apps on behalf of users and handling multi-step tasks. The report also mentions permissions and sandbox design, as well as warning screens requiring third-party AI tools to obtain user consent.
Why it matters to readers
If realized as reported, such agents would extend AI from answering questions to operating applications and devices on behalf of users; however, the materials currently describe development and plans, and the final functionality or release date cannot yet be confirmed.
For newcomers, the key is to distinguish between development plans and released products. The access permissions and sandbox mechanisms mentioned in the report relate to how users authorize AI to access devices and applications.
Risks and unknowns
The material is a paraphrase of Bloomberg’s report by PANews; no original Bloomberg article or recruitment page was provided, and the relevant information has not been independently verified in this material.
The specific date corresponding to “end of next year” is not specified in the materials; the release time and final features may still change.
The portion at the end of the material regarding the reactions of some AI companies was truncated, so their specific statements cannot be confirmed.
Related Developments
Loading event timeline…
Related concepts
AI agent: In the material, it refers to an AI system that can act on behalf of the user to perform operations and multi-step tasks.
This term is not in the glossary yet. Browse related concepts in the glossary.
View glossary →Sandbox mechanism: The material mentions it as part of the permissions and privacy design, but does not specify the exact implementation method.
This term is not in the glossary yet. Browse related concepts in the glossary.