Apple develops device-controlling AI agent amid increased hurdles for third-party rivals
Apple is entering the race to build AI agents capable of directly controlling user devices, following similar launches from Meta, OpenAI, Anthropic, and Google. A job listing shows the company is hiring a senior engineer to lead the effort, positioning the project to compete with Muse, Dots, Duet, and Grok Bots. Privacy is expected to sit at the core of the agent's design, relying on permission models and sandboxed test environments to keep actions protected.
The project remains in an early phase, though Bloomberg reports it could arrive as early as late next year. Apple's new CEO, John Ternus, is reportedly pressing software chief Craig Federighi to pursue a bolder AI strategy. That push comes after Apple recently rolled out generative AI features for Siri following repeated delays, internal rifts, and staff departures.
Even as it develops its own agent, Apple is raising new barriers for competing tools seeking access to its hardware. With macOS 27.3, users must now complete 14 authorization steps—up from four—to let third-party agents such as ChatGPT control their Macs. Apple says the expanded process is meant to protect users, while some AI companies and users view it as an effort to slow down competitors, according to Bloomberg.
Continue in the app — vote & join in ➔