Apple is developing an autonomous artificial intelligence (AI) agent to handle tasks on users’ behalf, joining a market contested by companies including OpenAI, Anthropic and Meta, Bloomberg reported.
Job advertisements posted on Apple’s website this week revealed the plans. One seeks a senior engineering manager to develop a proactive AI assistant. The advertisement says the system will be able to write code, browse the web, operate apps and use devices on users’ behalf.
The advertisement describes an AI agent that users can direct to operate a computer much as they would a capable assistant. The agent will be able to carry out multi-step tasks across apps, select suitable tools and coordinate sub-agents specialising in different tasks. Apple says safety and privacy must be central to the system’s design, with comprehensive permission controls, isolation mechanisms and transparency.
Apple plans to launch the service as early as the end of next year, but development is still at an early stage. An Apple spokesperson did not respond to Bloomberg’s request for comment.
Meanwhile, Apple is also tightening permissions for other AI agents to control Mac computers. In the forthcoming macOS 27.3 beta, users who want to let third-party agents such as ChatGPT operate their Macs will have to complete 14 steps, up from four. Apple says the move is intended to protect users, while some AI companies and users believe it could be an attempt to slow rivals’ progress.
Apple is still some way from launching a full-fledged AI agent, but has begun integrating its Siri digital assistant more deeply with third-party apps. The forthcoming iOS 27.3 will lay the groundwork for closer collaboration between Siri AI and external apps, allowing the assistant to book rides or make restaurant reservations on users’ behalf.
However, the updated system will still fall short of a full-fledged AI agent, which would need only limited instructions from users to carry out complex tasks autonomously in the background across multiple apps and services.