Anthropic will make Auto Mode the default in Claude Code
Anthropic will enable Auto Mode by default in Claude Code for Pro, Max, and Team subscribers starting August 14. Enterprise customers will still have to activate the feature themselves. Auto Mode allows the coding assistant to proceed without requesting approval for every operation. A classifier evaluates actions and asks for confirmation when it identifies a potentially dangerous or irreversible step.
In a controlled study involving 1,053 paid testers, Auto Mode detected 89 percent of dangerous commands, compared with 13.6 percent detected by human reviewers. Anthropic said teams using the feature created about 25 percent more pull requests. The company also reported that Auto Mode stopped a confidential-data upload and terminated about 2,000 processes that could have interfered with GPU training jobs.
Anthropic says the mode can also reduce prompt-injection risks. Trajectory Labs tested 72 attack scenarios ten times each, and reported no successful attacks against Claude’s current models—Fable 5, Opus 5, and Sonnet 5—in Auto Mode. The company still recommends human review for high-stakes production infrastructure changes, noting that the classifier lowers but does not remove risk.
Written by Sakhanda Wire AI from reporting by The Decoder. Facts, quotes and full context are in the original article below.