Anthropic is changing how Claude Code works. The company is turning auto mode on by default for users on Pro, Max, and Team plans. This means programmers will soon need to approve fewer actions when using the AI coding tool.
The change takes effect on Aug. 14. According to Techmeme, Anthropic says auto mode will be the default in Claude Code for these plans, claiming it is good enough at catching potential issues.
Why Anthropic is Making Auto Mode the Default in Claude Code
The decision comes from a simple problem: too many approval prompts. By default, Claude Code asks users for approval before running commands or modifying files. This keeps users safe, but it also means a lot of clicking "approve."
According to Anthropic Engineering, Claude Code users approve 93% of permission prompts. Over time, this leads to "approval fatigue," where people stop paying close attention to what they are approving.
Anthropic built classifiers to automate some of these decisions. The goal is to increase safety while reducing the number of times users have to click approve. The company explains what the system catches and what it misses.
How Auto Mode Works and What It Changes
Users currently have two ways to avoid approval fatigue. One option is a built-in sandbox where tools are isolated to prevent dangerous actions. The other is the --dangerously-skip-permissions flag, which skips approvals entirely.
Auto mode sits between these two options. It uses classifiers to decide which actions are safe to automate and which still need human approval. This approach aims to reduce the number of prompts without removing safety checks completely.
The company confirmed the rollout on its official channels. According to Claude's Threads post, auto mode is now the default in Claude Code for Pro, Max, and Team plans.
"Auto mode is now the default in Claude Code for Pro, Max, and Team plans." — Claude by Anthropic
What This Means for Developers Using Claude Code
For developers, this change means less time clicking through permission prompts. The 93% approval rate shows that most prompts are routine. Automating those decisions could speed up coding workflows significantly.
However, the remaining 7% of prompts are the ones that likely matter most. Anthropic's classifiers are designed to catch dangerous actions, but the company acknowledges the system has limits. Developers should understand what auto mode automates and what it still flags for review.
The change applies specifically to Pro, Max, and Team plans. Users on other plans will continue with the current permission system until further notice.
Our Take: A Smart Balance Between Speed and Safety
In our view, this is a sensible move from Anthropic. The company identified a real problem — approval fatigue — and built a targeted solution. Automating the 93% of routine approvals while keeping checks on the rest is a practical approach.
But developers should not become complacent. Auto mode reduces oversight, and that carries some risk. The system is designed to catch problems, but no automated classifier is perfect. Users should still review what auto mode approves, especially for critical projects.
The real test will come after Aug. 14, when developers start working with the new default. If auto mode handles routine tasks well and flags the right things, it could become a model for how AI coding tools handle permissions. If it misses something important, Anthropic will need to adjust quickly.
For now, the direction is clear: Anthropic wants Claude Code to require less human oversight. Whether that is a good thing depends on how well the safety systems hold up in real-world use.
