Claude Code Gets 'Auto Mode': AI Makes Permission Decisions on Your Behalf

Anthropic has launched a new "auto mode" for Claude Code, the coding tool that lets AI act autonomously on a user's behalf. The feature is designed to provide a safer middle ground between constant oversight and dangerous full autonomy.

Auto mode works by flagging and blocking potentially risky actions before they execute. If the agent is about to delete files, send sensitive data, or execute hidden instructions, the system stops and either gives the agent another chance or asks the user to intervene.

Anthropic frames this as a direct response to a real dilemma in AI agents: full autonomy is efficient but risky, while constant user confirmation is safe but slow. Auto mode is designed to find the middle ground.

Currently, the feature is available as a research preview for Team plan users. Anthropic says access will expand to Enterprise and API users "in the coming days." The company warns the tool is experimental and "doesn't eliminate" risk entirely, recommending developers use it in isolated environments.

This is a direct follow-up to the larger "computer use" launch earlier this week, in which Claude Code gained the ability to open apps, use browsers, and control mouse and keyboard on a user's Mac.