Anthropic is moving Claude Code into a more autonomous era by making its auto mode the default for Pro, Max, and Team accounts starting August 14. The update is designed to reduce the need for constant approval prompts while keeping key safeguards in place.
In auto mode, the coding assistant continues working without stopping for every step, unless it detects an action that is irreversible, destructive, or outside the user's environment. Anthropic says the feature was first tested in March as a way to combine speed with control.
The company says its internal testing suggests the system can outperform manual review in some cases. Among 1,053 paid testers, auto mode identified 89% of harmful actions, while human review caught 13.6%. Anthropic also noted that users approve permission prompts at a very high rate, which can make manual checks less effective over time.
Claude Code lead Boris Cherny said the team has been using auto mode exclusively for months, describing it as a workflow that feels natural for modern software development. Anthropic has also added protections such as prompt injection screening and customizable hard-deny rules to help block risky behavior, including data leakage attempts.
The shift reflects a broader trend in AI tools: moving from assisted interaction toward smarter, more independent execution. If adopted widely, this approach could reshape how developers build, test, and ship software in the years ahead.
