Anthropic releases safer Claude Code 'auto mode' to avoid mass file deletions and other AI snafus

Anthropic began previewing auto mode in Claude Code, anthropic began previewing auto mode inside Claude Code.

Auto mode functions as an intermediate option between per-action approval and fully skipping permissions auto mode operates between default approval behavior and a skip-permissions command.

Classifier guides Claude to allow safe actions and to redirect risky actions a classifier system lets Claude perform actions it deems safe and redirects it when actions appear risky.

Anthropic designed auto mode to reduce mass file deletions, sensitive data extraction, and malicious code execution, anthropic designed the system to reduce mass file deletions, sensitive data extraction, and execution of malicious code.

Anthropic statement warned the classifier may still allow some risky actions if intent is ambiguous or context is insufficient, anthropic warned the classifier may still allow some risky actions when intent is ambiguous or context is insufficient.