Anthropic to make Claude's Auto Mode default for Pro, Max and Team users

Anthropic will make Auto Mode the default setting for new sessions in most Claude apps starting next week, the company said. The change applies to Pro, Max and Team plan users, according to aitimes.com.

Auto Mode automatically checks requests to execute tools, letting through those judged safe and blocking those judged risky. Blocked tasks can be retried by another route or sent to the user for approval, and users can switch back to manual approval at any time.

Anthropic said Claude Code users approved 97% of permission requests. In a coding experiment with 1,053 paid professional testers, humans blocked 13.6% of clearly risky commands inserted into approval requests, while Auto Mode blocked 89% of the same commands. Human block rates fell to about 5% after more than 50 prior approvals, while Auto Mode's rate did not vary with session length.

The checks cover hard-to-reverse changes, destructive commands and data sent outside the system. Anthropic cited blocked cases including an attempt to post a report to a public code-sharing site and an attempt to grant cloud admin rights beyond the requested scope.

Enterprise plan administrators will be notified by email before Auto Mode is applied by default. Anthropic recommends that users manually review operations that make significant changes to production infrastructure.

In August 2026, Anthropic made Auto Mode the default in its coding tool Claude Code.