Anthropic adds auto mode to Claude Code for safer unattended work

Anthropic has launched an automatic mode for Claude Code, its AI-based development tool that lets the model carry out tasks on the user's behalf. The new auto mode tackles one of the central problems of autonomous agents: how to give an AI enough freedom to work effectively without risking data loss, information leaks or unwanted commands.

Before auto mode, Claude Code users faced an awkward choice. They could review every action of the agent manually, which is reliable but slow and tiring. Or they could grant the model broad permissions and stay out of the way, which is fast but risky. Claude Code can already work on its own: delete files, send data, run system commands. That makes it powerful, but also potentially dangerous, especially if the agent runs into hidden instructions in third-party code or malicious data.

Auto mode is designed as a third path. The new feature analyzes the agent's upcoming actions, assesses their potential risk and blocks dangerous operations before they are executed. The agent can then suggest an alternative, safer way to reach the same goal, rather than simply refusing the task. Anthropic positions this as a balance between constant supervision and a dangerous level of autonomy, which the company considers unacceptable for a broad audience of users.

In essence, auto mode is a built-in layer of self-censorship for the agent. Instead of asking the user for permission at every step or acting without any checks, Claude Code now weighs the risks itself in real time. This resembles the approach used in industrial safety systems: don't block every non-standard action, but evaluate the context and respond to real threats.

The problem that auto mode is meant to solve is well known to developers. AI agents are vulnerable to so-called prompt injection attacks, in which malicious instructions are hidden in the data, documents or third-party library code the agent processes. The agent dutifully follows them, unaware of the threat. In addition, broad permissions in autonomous mode can lead to unwanted outcomes even without malicious intent: deleting needed files, sending confidential data to external services, or running destructive operations on databases.

The launch of auto mode is part of a broader industry trend. As AI agents become more autonomous, controllability and safety move to the forefront for all key players in the market. Companies are looking for ways to give users powerful tools without creating new vulnerabilities. Anthropic is clearly betting that trust in Claude Code will grow along with its reliability and predictable behavior.

For developers who use Claude Code in real projects, auto mode is a practical workflow improvement. There is no longer a painful choice between speed and safety. The agent takes on part of the decisions, but only those that can be delegated, leaving critical and irreversible actions under the user's control. It is a step toward AI tools that can be trusted in a real working environment.