Anthropic is turning Claude Code’s auto mode on by default

Programming with Claude Code is about to become a more autonomous experience. Starting August 14, Anthropic is shifting the default setting for Pro, Max, and Team accounts to "auto mode," significantly reducing the frequency of manual oversight required during development workflows.

A Shift Toward Autonomy

First introduced as a pilot in March, auto mode was designed to strike a balance between developer velocity and system control. Under this new default configuration, the AI will execute tasks continuously rather than pausing for human approval at every turn. The system will only halt to request permission if it identifies an action that is:

• Irreversible
• Destructive
• Targeted outside the local environment

Safety Through Automation

While the move toward automation might raise concerns regarding oversight, Anthropic reports that its internal testing suggests auto mode is actually more secure than manual intervention. In a study involving 1,053 paid users, the automated system successfully flagged 89% of potentially harmful actions, whereas human reviewers caught only 13.6%.

"The team and I use Auto mode exclusively, and have been for many months. I couldn’t imagine going back to permission prompts!" — Boris Cherny, Head of Claude Code.

The company attributes this disparity to "approval fatigue," noting that users typically greenlight 97% of permission prompts out of habit. To bolster this transition, Anthropic has implemented robust safety measures, including prompt injection screening and customizable hard deny rules designed to prevent unauthorized data exfiltration.