Archive — US — techcrunch
Anthropic is turning Claude Code’s auto mode on by default
Anthropic is enabling auto mode as the default for its Claude Code programming tool across Pro, Max, and Team accounts starting August 14. This mode minimizes human oversight, operates safely with high error-catching rates, and eliminates frequent permission prompts. Enhanced safety features are also introduced to prevent potential issues like data leaks.
Key points
- Beginning August 14, auto mode will be default on Claude Code for Pro, Max, and Team accounts.
- Auto mode reduces human oversight by proceeding with actions unless deemed irreversible or destructive.
- In tests, auto mode caught 89% of harmful actions, compared to only 13.6% by manual review.
- Boris Cherny, Claude Code Head, advocates for auto mode, indicating it improves user experience.
- New safety features include prompt injection screening and customizable denial rules to safeguard against data exfiltration.
Why it matters
- The move reflects ongoing trends in AI tools seeking to enhance automation in programming tasks.
- By 2025, increased automation in IT is expected to reduce human error and improve efficiency, aligning with Claude's approach.
- The emphasis on safety features highlights growing concerns about AI misuse in sensitive environments.
Anthropic's decision to default to auto mode in Claude Code represents a significant advancement in programming efficiency, aligning with broader AI trends prioritizing automation and safety in technology.