Apple Forms AI Agent Team as macOS Raises the Authorization Bar for Third-Party Agents
Apple is assembling an AI agent team with the goal of enabling agents to write code, browse the web, operate apps, and control devices on users' behalf. However, the project remains at an early stage, and its launch date has yet to be determined. The forthcoming macOS 27.3 beta adds permission warnings for third-party AI agents, and reports say the number of steps ChatGPT users must take to enable third-party computer-control tools has increased from four to 14.
Why read this
AI score 65
Job postings and macOS permission changes together reveal Apple's agent strategy, offering a way to assess the potential impact of the platform's in-house products on third-party tools.
Topics
Body · AI translation
Cailian Press Financial News Agency
Open App
Apple Forms AI Agent Team as macOS Sets a Soft Barrier for Rivals
STAR Market Daily, Song Ziqiao; Editor: Ji Sheng; 2026-11-20 10:14
① Apple wants its AI agents to write code, browse the web, operate applications, and control devices on users’ behalf; ② Reports say the project is still at an early stage. Apple has considered launching the related service as early as the end of 2027, but the timeline has yet to be determined; ③ If Apple launches its in-house product, it may become more difficult for competitors to use AI agents to control Mac computers.
STAR Market Daily, November 20 (Editor: Song Ziqiao) — On November 20 local time, technology journalist Gurman reported that Apple is developing its own AI agents.
Gurman’s report
On November 18 local time, job listings posted on Apple’s official website revealed the plans. One listing showed that Apple is recruiting a leader for its AI agent team—a Senior Engineering Manager who will serve as the team’s senior leader. The position is titled “Agentic AI-Proactive.”
Image source: Apple’s official website
“Imagine an AI agent that can use a computer like a capable person,” the job listing reads. “It can browse the web, work across apps, run commands, write and debug code, invoke services, and choose the right tool at every step.” Apple also placed particular emphasis on privacy protection, saying that agents capable of running code, using user credentials, and taking action on users’ behalf require rigorous permission management, sandboxing, and operational transparency. “It does all of this privately, with the polish people have come to expect from Apple products.”
Under “Job Responsibilities,” Apple states:
“As the team’s senior leader, you will have full responsibility for the technical direction and delivery of the agent platform and the various AI capabilities built on it. You will lead and develop a team of managers and engineers while building a talent pipeline in AI and systems to lay the groundwork for our future product delivery. You will need to work closely with multiple partner teams, including Siri, UI frameworks, and AFM, with the core objective of integrating capabilities across teams to build agents that are fully featured, stable, reliable, secure, and compliant.”
According to the report, the project remains at an early stage. Apple has considered launching the related service as early as the end of 2027, but the timeline has not yet been finalized. Apple declined to comment.
Gurman said Apple’s new CEO, John Ternus, has been pressing software engineering chief Craig Federighi to accelerate AI development and adopt a more aggressive strategy. Autonomous AI agents are seen as an important part of Apple’s efforts to advance this strategy.
OpenAI, Anthropic, Google, Meta, and other companies have each launched their own agents. These products are highly popular on the App Store. As of Thursday local time, Meta’s Muse was the second-ranked free app, behind only TikTok.
Gurman believes that if Apple launches its in-house product, it will become more difficult for competitors to use AI agents to control Mac computers.
Apple has already begun adjusting macOS permission management for third-party applications. The upcoming macOS27.3 beta adds permission warnings for third-party AI agents, requiring users to reconfirm when granting such applications permission to control a Mac. The changes increase the number of steps ChatGPT users must complete to enable third-party computer-control tools from four to 14. Apple, however, has described the changes as measures to strengthen the protection of user data.
China Galaxy Securities recently said in a research report that global Agent applications currently exhibit three major trends. First, demand for task execution is growing rapidly: global execution-oriented Agent Token usage reached 59.96 billion in October, approximately 7.2 times the volume under the human-execution model. Second, enterprise production deployment has accelerated significantly: in 2026, 71% of surveyed U.S. enterprises had deployed Agents in production, up 31 percentage points from the previous year. Third, divergence among industry applications has begun to emerge: the rate at which pilots moved into production exceeded 50% in banking and insurance, software and internet, and telecommunications, while manufacturing, healthcare, and government departments lagged behind.
The institution said that data governance, system integration, and the ability to reshape business processes will become key factors affecting the large-scale deployment of Agents. Industries with stronger digital foundations are expected to be the first to establish an advantage in adoption.
A China Securities research report said that flagship models and agent products continue to be upgraded and maintained a positive outlook on leading platforms that combine models, cloud resources, and application entry points.
Artificial Intelligence
Apple Supply Chain
Cailian Press Statement: This article is for reference only and does not constitute investment advice. Investors act on this information at their own risk.
Business Cooperation
Popular Unlocks
[Market Trend Research Report · Company] After entering the high-end optical communications testing sector, the company’s performance grew by nearly 300% in the first three quarters. Combined with the global expansion of manufacturing capacity in its core business, this is expected to jointly open up a second growth curve.
The company has cultivated its traditional core business for 30 years, establishing it as the foundation of its operations. In recent years, it has entered optical communications testing fields such as high-speed optical modules and silicon photonics chips through acquisitions. The acquired business has joined the supply chains of leading global optical communications manufacturers and continues to make breakthroughs with new high-end testing products, potentially opening up a second growth curve.
[Intraday Tip] Global sales surge nearly 150% year on year; AI is a key driver of industry growth, and amid the trend toward domestic substitution, the sector is expected to maintain a high level of prosperity. A subsidiary of this company has already secured orders from multiple customers
Global sales surged nearly 150% year on year. AI is a key driver of industry growth, and amid the trend toward domestic substitution, the sector is expected to maintain a high level of prosperity. A subsidiary of this company has already secured orders from multiple customers, while another company recorded increases in both product sales volume and prices in the third quarter and further expanded its business scale through an acquisition.
[Mastering ETFs] Prices rise by more than 20% at most as overall capacity in the industry remains tight; institutions say several companies are expected to make further progress with next-generation technologies, and the related sector may continue to see increases in both volume and prices
① Prices have risen by more than 20% at most, while overall capacity in the industry remains tight. Institutions say that looking ahead, several companies are expected to make further progress with next-generation technologies, and the related sector may continue to see increases in both volume and prices; ② Institution: Dividend assets are relatively expensive, making how to optimize current dividend-asset allocations a key issue.
Related Reading
Moving Beyond Benchmark Races: China's AI Industry Shifts Competition Toward “Token Output Efficiency” | Industry Watch
STAR Market Daily reporters Tian Ye and Huang Xinyi, 2026-11-20 09:00
① Investment in domestic AI computing infrastructure continues to grow rapidly, with large numbers of intelligent-computing centers and domestically produced accelerator cards coming online, substantially expanding the total supply of computing power. ② The industry clearly exhibits a pattern of “high volume but low effectiveness”: ecosystems based on different technological approaches are fragmented, adapting chips to models is difficult, and computing capacity has a relatively high idle rate, making it difficult to convert the expansion of hardware capacity into actual business output.
Nvidia's Bleeding Continues: Market Value Plunges by More Than RMB 4 Trillion in Just Six Trading Days
Cailian Press, Bian Chun, 2026-11-20 08:12
① Nvidia shares fell again on Thursday, wiping out a cumulative $669 billion (approximately RMB 4.48 trillion) in the company's market value over the past six trading days; ② Only 20 companies worldwide have a market capitalization of at least $669 billion; ③ Some analysts believe Nvidia's share-price decline is not yet over, as multiple factors are weighing on the stock.
Will the AI “Memory Shortage” Worsen? Micron President: Supply and Demand Will Be Even Tighter in 2027 Than This Year!
Cailian Press, Huang Junzhi, 2026-11-20 08:08
① Micron Technology executives said the shortage of memory-chip supplies is unlikely to ease in the short term, and supply and demand will be even tighter in 2027 than in 2026; ② The company has already presold 75% of its projected 2027 capacity and signed 26 multiyear customer agreements involving a total of $32 billion in prepayments and funding commitments.
AI Trends