Anthropic is taking another step toward more autonomous AI-assisted software development by making Claude Code’s Auto mode the default for Pro, Max and Team accounts starting August 14.
In Auto mode, Claude Code can proceed without asking for human approval at every step. It will still stop when an action is considered irreversible, destructive, or outside the user's environment.
Anthropic says testing showed Auto mode caught 89% of harmful actions, compared with 13.6% through human review. The company also noted that users approved 97% of permission prompts, suggesting that repeated manual approvals can become routine rather than meaningful security checks.
The company is also adding safeguards including prompt-injection screening and customizable hard-deny rules, including protections against data exfiltration.
This signals a major shift in AI coding: developers are moving from AI that suggests code toward AI that can increasingly execute development tasks autonomously.
For businesses, this could dramatically improve developer productivity—but it also makes AI governance, permissions, monitoring and security controls increasingly important.
