How exactly are we making sure the commands the model executes are not malicious or malformed?
I just had GLM 5.2 search with find command in my root directory..
Not a privacy issue persé, but it's not very logical to do that, and if it is capable of doing that it might do worse things on my /
How exactly are we making sure the commands the model executes are not malicious or malformed?
I just had GLM 5.2 search with find command in my root directory..
Not a privacy issue persé, but it's not very logical to do that, and if it is capable of doing that it might do worse things on my
/