Anthropic announced last week that it is changing the way how Claude Code works. The company revealed that its Auto mode will become the default setting for Pro, Max and Team users starting August 14.
This simply means that developers using the Claude Code will not have to manually approve every action taken by the AI. Rather, Claude will continue working on its own unless it detects that an action could be irreversible, destructive or outside the user’s environment.
Claude Code Will Need Less Human Approval
Anthropic first introduced Auto mode as a test in March. This feature was designed to give developers a way to let Claude work faster while still maintaining safety controls. Under the new default setting, Claude Code will automatically continue with tasks instead of repeatedly stopping to ask the user for permission.
However, the AI will stop when it identifies actions that may cause serious damage or move beyond the user’s working environment. This could make Claude Code particularly useful for large programming tasks that require the AI to work for longer periods.
The American tech giant claims that Auto mode performed better than manual approval during its testing. Anthropic tested the feature with 1,053 paid users and said Auto mode detected 89 per cent of harmful actions. In comparison, human reviewers caught only 13.6 per cent of such actions.
Anthropic has also stated that its Auto mode can improve productivity. Among the Teams and Enterprise users, Anthropic said those using Auto mode shipped around 25 per cent more pull requests.
According to Anthropic, companies including Adobe, Nuro, Gusto and Garner Health are already using Auto mode as their production default.
Boris Cherny, Head of Claude Code, said the team behind the product has already been using the feature extensively.
In a post on X, Cherny said, “The team and I use Auto mode exclusively, and have been for many months. I couldn’t imagine going back to permission prompts!”
