The Anthropist gives Claude Codd more control, but keeps him on a leash

For developers using AI, “vital programming” today is to monitor every action or risk letting the model run unchecked. Anthropic says its latest update to Claude aims to eliminate that choice by letting the AI ​​decide which actions are safe to take on its own — with some limitations.

The move reflects a broader shift across the industry, as AI tools are increasingly designed to work without waiting for human approval. The challenge is to balance speed and control: too many guardrails slow things down, while too few can make systems risky and unpredictable. Anthropic’s new Auto Mode, which is now in research preview — meaning it’s available for testing but not a finished product yet — is the latest attempt to thread that needle.

Automated mode uses AI protections to review each action before running it, checking for risky behavior that the user did not request and signs of spot injection — a type of attack where malicious instructions are hidden in content processed by the AI, triggering unintended actions. Any safe actions will be followed automatically, while dangerous actions will be blocked.

It’s essentially an extension of Claude Code’s existing “dangerously-skip-permissions” command, which hands over all decision-making to the AI, but with an added layer of security on top.

This feature relies on a wave of independent programming tools from companies like GitHub and OpenAI, which can perform tasks on the developer’s behalf. But it takes it a step further by shifting the decision about when to ask permission from the user to the AI ​​itself.

Anthropic hasn’t detailed the specific criteria its security layer uses to differentiate safe actions from risky ones — something developers will likely want to better understand before the feature is widely adopted. (TechCrunch has reached out to the company for more information on this front.)

Automated mode follows Anthropic’s launch of Claude Code Review, an automated code reviewer designed to catch bugs before they reach the code base, and Dispatch for Cowork, which allows users to submit tasks to AI agents to handle the work for them.

Automatic mode will be rolled out to Enterprise and API users in the coming days. The company says it currently only works with Claude Sonnet 4.6 and Opus 4.6, and recommends using the new feature in “isolated environments” — sandbox settings that are kept separate from production systems, limiting potential damage if something goes wrong.