ChatGPT's Private Keys

2 pointsjzombie3 comments

Using ChatGPT4 plus, I've asked it to write a program that lists the contents of the current directory.

Inside, there's some files with the name: kernel-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx.json

If you then ask it to read one of them, it will show you some keys, ports, etc.

I wonder if this can lead to other things.

Why do you think what a large language model is telling you is something that is being executed?

The Python it writes runs in a sandbox environment with its own containerized filesyatem. That's been obvious to me for a while as I have seen it generate temporary files when uploading things like images to scan.

This does not imply that any large language model executes the code it writes... I get it, they don't. I build models myself and have a fairly good understanding of how they work. But ChatGPT-4 does provide the ability to generate and run code in its own sandbox environment, outside of the model itself.

For instance, if I were using someone else's custom GPT, I wonder if it would be using my particular container system, or someone else's.

Reply on news.ycombinator.com