Auto mode is now the default in Claude Code for Pro, Max, and Team plans
Claude Code auto mode becomes default, blocking 89% of harmful actions vs humans' 13.6%
“In this evaluation, none of the 720 attack attempts succeeded against Claude Fable 5, Opus 5, or Sonnet 5 running auto mode.”
Anthropic is making auto mode the default in Claude Code for Pro, Max, and Team plans starting August 14th, citing a controlled study where auto mode blocked 89% of harmful actions versus humans refusing only 13.6%. A third-party evaluation by Trajectory Labs found zero successful prompt injection attacks across 720 attempts against the latest Claude models in auto mode, a significant claim about solving the 'lethal trifecta' of agentic AI security risks. This represents a major shift in how Anthropic is positioning agentic AI safety — moving from human-in-the-loop confirmation to model-driven autonomous safety enforcement.