In testing, auto mode demonstrated a higher safety rate, identifying 89% of harmful actions compared to just 13.6% with manual review. The company is also enhancing safety features to mitigate risks such as data exfiltration.
With auto mode's impressive safety performance, users should monitor how quickly Anthropic rolls out these features across its platforms. Watch for updates on customizable safety settings, which could empower users to tailor Claude Code's responses to their specific environments.