Anthropic is shifting Claude Code's auto mode from an optional test feature to the default setting for Pro, Max, and Team accounts, starting August 14. The change means the coding assistant will no longer pause for human approval at every step, instead proceeding unless an action is flagged as irreversible, destructive, or outside the user's environment.

The company first piloted auto mode in March, framing it as a balance between speed and control. In its Friday announcement, Anthropic reported that internal testing with 1,053 paid participants showed auto mode caught 89% of harmful actions, compared to just 13.6% for manual review. Anthropic suggested that manual review often becomes habitual, noting users approve 97% of permission prompts in Claude Code.

Claude Code Head Boris Cherny said on X that he and his team have used auto mode exclusively for months and couldn't imagine returning to permission prompts. Anthropic also highlighted recent safety additions, including prompt injection screening and customizable hard deny rules aimed at preventing data exfiltration.

Source: TechCrunch