Claude Code Flips to Auto Mode: AI Blocks 89% of Dangerous Commands vs 14% by Humans
Justgototheeffinmoon · reddit · 2026-08-09
Anthropic has flipped Claude Code to Auto Mode by default. An internal study of 1,053 paid testers showed the AI classifier blocked 89% of dangerous commands, compared to just 13.6% by humans. Human performance reportedly dropped to around 5% after 50 prompts due to approval fatigue.
Production data indicates manually-approved sessions caused unintended harm twice as often as Auto Mode. Additionally, Team and Enterprise customers shipping roughly 25% more pull requests with Auto Mode, and Anthropic will stop billing for the extra tokens consumed by the classifier.
More from coding & agent
- DIRECT: retrieval-based next-question suggestions for conversational agents — burkov · 2026-10-16
- Claude Code and Codex now autocomplete the user's next action — tokenbender · 2026-10-16
- Hermes Agent ships official Reverse Engineer Anything plugin with one-command setup — Teknium · 2026-10-16
- MLExpert: 35B MoE Agent Trained to Autonomously Handle Machine Learning Engineering — Shangheng Du · 2026-10-16
- Cowork-Gym: Alibaba Synthesizes 64.9K Agent Work Environments, Boosting Qwen by 8.7 Points — Anni Zou · 2026-10-16
- MMSearch-OPD: Joint RL and On-Policy Distillation for Long-Horizon Multimodal Search Agents — PekingUniversity · 2026-10-16