Posted by sbehere 2 hours ago
A lot of the discussion about these long sessions where agents are left to operate autonomously feels like listening to the increasingly drunk guy at the bar who says "I ran IT at that Fortune 100 place for a decade and we never had a single problem using a short but loose rule set for the firewall until last week someone destroyed our entire business in 27 minutes".
I have been using Codex with auto-approve mode for a couple months and haven't had a single incident (or at least haven't noticed). Maybe as capabilities get better and better and they are less likely to do something dumb like wiping ~/, we can just trust them?
I guess this argument works unless we worry about agents doing something out of malice instead of stupidity.
Imo these explicit tool-level permissions are really just a bandaid for bad sandboxing. Just be aware of where you are running your agent and what data is at risk of being destroyed or compromised. Assume that arbitrary code can run at any time and be prepared to recover from that.
I've been running both in yolo mode and haven't had a single incident.
---
None of this is really about figuring out how to protect people's drives, in my opinion. The real issue is a deep session where Ada is using Claude Code to get a refund and at some point the system "exploits" the merchant's api without any malicious intent.
In my opinion, this is a complex thing because it's more about reward hacking and an already aligned model thinking it's doing the right thing. So another aligned model monitoring actions might just falter via inheritance. You could imagine they account for proper layering/intent+action-isolation in their auto mode architecture.
How do you know for sure?
However, I don't give it long running tasks unsupervised, and I do interrupt it from time to time to give suggestions.
What’s the best way to sandbox Claude on macOS without it being a huge hassle?