Skip to main content
Back to News Hub
🟢TechCrunch AI
August 9, 2026
Claude Code

Anthropic is turning Claude Code's auto mode on by default

Overview

Anthropic is making auto mode the default setting for Claude Code across Pro, Max, and Team accounts starting on August 14. This feature allows the system to execute tasks without requiring prompt approvals at every step, unless an action is destructive, irreversible, or directed outside the user's environment. Internal testing with 1,053 paid testers showed auto mode identified 89% of harmful actions compared to 13.6% caught by manual human review.

Key Takeaways

  • Anthropic announced that it will activate auto mode by default for Claude Code starting August 14 for subscribers on Pro, Max, and Team plans.

    First introduced as a test version in March, auto mode reduces the need for constant human oversight during coding sessions.

  • The decision to make auto mode standard follows evaluation data regarding safety and human oversight patterns.

    In a study involving 1,053 paid testers, auto mode intercepted 89% of dangerous actions, whereas human reviewers detected only 13.6%.

  • Boris Cherny, the head of Claude Code, noted in a post on X that his team uses auto mode exclusively.

    To address security concerns such as data exfiltration, Anthropic has incorporated extra protective mechanisms, including prompt injection screening and customizable hard deny rules.

  • Anthropic will enable auto mode by default for Claude Code Pro, Max, and Team users on August 14.

    In a study of 1,053 paid testers, auto mode detected 89% of harmful actions while human reviewers caught only 13.6%.

  • Anthropic is integrating prompt injection screening and customizable hard deny rules to protect against risks like data exfiltration.

Stats & Key Facts

  • #Internal testing with 1,053 paid testers showed auto mode identified 89% of harmful actions compared to 13.6% caught by manual human review.
  • #Anthropic announced that it will activate auto mode by default for Claude Code starting August 14 for subscribers on Pro, Max, and Team plans.
  • #In a study involving 1,053 paid testers, auto mode intercepted 89% of dangerous actions, whereas human reviewers detected only 13.6%.
  • #Data revealed that manual approval tends to become routine, with users agreeing to 97% of permission prompts.

Anthropic announced that it will activate auto mode by default for Claude Code starting August 14 for subscribers on Pro, Max, and Team plans. First introduced as a test version in March, auto mode reduces the need for constant human oversight during coding sessions. Instead of pausing for explicit user approval at every step, the system continues working autonomously unless it encounters an action that is "irreversible, destructive, or aimed outside your environment."

The decision to make auto mode standard follows evaluation data regarding safety and human oversight patterns. In a study involving 1,053 paid testers, auto mode intercepted 89% of dangerous actions, whereas human reviewers detected only 13.6%. Data revealed that manual approval tends to become routine, with users agreeing to 97% of permission prompts.

Boris Cherny, the head of Claude Code, noted in a post on X that his team uses auto mode exclusively. To address security concerns such as data exfiltration, Anthropic has incorporated extra protective mechanisms, including prompt injection screening and customizable hard deny rules. For those studying real-world AI deployment, this transition demonstrates how automated safety checks can be more effective than human monitoring when users suffer from permission prompt fatigue.

For more details please read the original article at TechCrunch AI.

Continue Learning

Comments

Comments appear only after moderation. Your email identifies your submission to the moderator and is never displayed here.

No approved comments yet.

Originally published by TechCrunch AI
Read the original