PyPI Suspends New User and Project Creation in Wake of Malware Campaign
blog.phylum.ioIt is scary. Here they caught the culprit quickly, but if they've been careful, I think they would be able to stay under the radar and still infect a fair few systems.
As a person who regularly runs pip install on my main desktop, I am definitely worried about arbitrary code execution that happens when you pip install. Sure I can run everything inside the container, but given that I do most of my work in python, I think that is too restrictive...
Yeah, the broad campaign makes it extremely noticeable. There are active campaigns right now that don't take this approach. Singular packages with novel malicious payloads.
> As a person who regularly runs pip install on my main desktop, where I am worried about arbitrary code execution that happens when you pip install.
We've open-sourced a sandbox and wrapped the Phylum CLI with it so you can do something like `phylum pip install <pkgName>,` it'll check our API first for known malware, then if it appears clean, will perform the installation in the sandbox. You can specify what the sandbox is allowed to touch in a TOML file.
This is great. Is there something for crates.io?
Does the safety-oriented Rust community do this _automatically_?
Honest question: Is this unique to python? or can we expect this in Go, Rust, vcpkg, conan, etc?
No, this is not unique to Python or PyPI. I'm one of the co-founders @ Phylum. We've tracked campaigns across Crates.io, Nuget, npm, PyPi, etc.
This one gained more traction: https://news.ycombinator.com/item?id=39856756
run littlesnitch or something similar to notice and prevent egress attempts. for now it seems the only effective defense.
hopefully somebody builds a disk snitch. would love to buy that.