Auto mode is now the default in Claude Code

Anthropic just made Auto mode the default in Claude Code for paid Pro, Max, and Team users starting August 14, which basically means the coding assistant will be allowed to keep working on its own for longer stretches instead of constantly stopping to ask for permission. The company says this setup can even catch more risky commands than manual review in testing, and it says the tiny extra usage cost behind the feature is no longer being charged on those plans. Sounds tidy, right? The comments immediately turned it into a safety drama.

One camp is basically cheering, saying this should have happened sooner because the nonstop permission pop-ups were driving people up the wall. As one commenter put it, the tool was prompting users "incessantly", so making the smoother setting the default feels like common sense. But the other camp is waving giant red flags and yelling, in plain English, "cool, but lock this thing in a box first." One person called it an excellent moment to review sandboxing — a way of fencing software in so it can’t run wild — while another casually dropped that they’ve been using the system with "dangerously-skip-permissions" inside Docker, basically a sealed container, and would never run it loose.

And because no internet discussion can stay on one topic, someone popped in with a side-quest complaint: "And Fable is gone too." Then came the classic Hacker News mini-drama opener: "This is a duplicate." So yes, the launch sparked the full buffet — safety fears, convenience stans, product grieving, and hall-monitor energy all at once.