Anthropic says it will make auto mode the default setting for Claude Code users on Pro, Max and Team accounts from August 14.
The company first introduced a test version of auto mode in March, describing it as a way to balance speed and control for programming tasks.
In auto mode, Claude Code proceeds without seeking approval at every step unless it identifies an action as irreversible, destructive or directed outside the user’s environment.
Anthropic said testing showed auto mode identified 89% of harmful actions in a study involving 1,053 paid testers. Human review identified 13.6% of harmful actions, according to the company.
It added that manual review can become habitual, with users approving 97% of permission prompts in Claude Code.
Claude Code head Boris Cherny said he and his team have used auto mode exclusively for several months.
Anthropic also said it has added safety measures including prompt-injection screening and customisable hard-deny rules designed to prevent actions such as data exfiltration.
