Anthropic enables Claude to execute computer tasks in the background

Anthropic has updated its desktop assistant, enabling Claude to perform computer operations in the background. According to a post by Anthropic on X, the new feature allows Claude Cowork and Claude Code to launch applications, type text, and navigate software without requiring active supervision on screen. The capability is currently available in preview mode for paid subscribers on macOS.

This update marks a notable evolution in how people interact with artificial intelligence. Rather than just answering questions in a chat window, the system acts as an agent that handles administrative chores quietly behind the scenes. Users can delegate repetitive desktop tasks to the software and continue working on unrelated assignments at the same time.

However, allowing an automated tool to control a personal device out of sight raises key safety and practical questions. It is still unclear how reliably the assistant handles unexpected software bugs, misclicks, or security prompts when running unattended. As background automation becomes more common, tech companies will likely face growing pressure to guarantee that these digital agents do not accidentally erase files or leak sensitive user data.