Claude Desktop starts a VM silently with no way to stop it
Claude Desktop (Anthropic's AI app) reportedly launches a virtual machine (VM) on its own, without the user's knowledge. Once it starts, there is no built-in way to stop or shut it down. The issue has been filed publicly on Anthropic's GitHub.
A virtual machine (VM) is essentially a mini computer running inside your real computer, used to safely run code or isolated tasks. Claude Desktop appears to spin one up automatically — likely to handle code execution — but gives users no button, menu, or command to turn it off afterward.
A VM running in the background silently consumes memory and CPU even when you think Claude is idle. This is a reported bug, not an intended feature, and it is currently open on Anthropic's official GitHub issue tracker. If you use Claude Desktop, it is worth checking your system's resource usage after a session to see if anything unexpected is still running.
Key points
- Claude Desktop launches a VM automatically, without asking the user
- There is no provided way to stop or close the VM once it starts
- The background VM can quietly consume memory and CPU resources
- The issue is publicly filed as a bug on Anthropic's GitHub
- Users should check system resource usage after using Claude Desktop
Quick term guide
- Claude Desktop
- Anthropic's computer app for using Claude outside the browser.
- virtual machine
- A software-based computer that runs inside your real computer, letting you run a different operating system (like Linux) on your Mac.
- edge
- Servers closer to users, so apps can respond faster.
- compute
- The server power and chips needed to run AI systems.
- memory
- A ChatGPT feature that lets it use details from past chats in future chats.
- usage
- How much of a tool or service you have used.
- session
- A continuous period of interaction between a user and a computer program.
- sources
- Evidence showing where a piece of information came from.