Anthropic makes Claude Code auto mode default for Pro, Max, Team

Anthropic is shifting Claude Code, its AI coding assistant, to auto mode by default for Pro, Max, and Team accounts starting August 14. Previously tested since March, auto mode eliminates step-by-step permission prompts and instead lets the agent act autonomously—unless an action is flagged as irreversible, destructive, or aimed outside the project environment. The company claims this actually improves safety. In a study with 1,053 paid testers, auto mode caught 89% of harmful actions, while human review caught only 13.6%. Anthropic attributes the gap to “habitual” manual review, noting users approve 97% of permission prompts in Claude Code anyway.

Claude Code Head Boris Cherny said on X that the team has used auto mode exclusively for months and “couldn’t imagine going back to permission prompts.” Anthropic also introduced new guardrails: prompt injection screening to neutralize hidden malicious instructions, and customizable hard deny rules that block specific high-risk operations like data exfiltration. For developers, this marks a major shift toward trusting AI agents in production workflows—fewer interruptions, faster iteration, but also new considerations around oversight and security boundaries.