Google released a 1MB AI model that quickly catches malware disguised as PDFs.
AlphaSignal
Visit on X@AlphaSignalAI17K followers
13,216 views
Google released a 1MB AI model that quickly catches malware disguised as PDFs.
Magika is an open source file detection system.
It doesn't trust file extensions.
Instead, it reads the actual content to figure out what a file really is.
That means malware disguised as a PDF gets caught.
Hidden scripts inside images get flagged. Fake extensions don't fool it.
It runs on a single CPU and classifies files in about 5ms.
It supports 200+ file types and hits ~99% accuracy across ~100M training samples.
It works as:
> A Rust command line tool
> A Python library via pip
> JavaScript and Go bindings
Google already uses it across Gmail, Drive, and Safe Browsing.
Now anyone can run it locally, scan entire directories, and build it into their own security pipelines.
4 Comments
Comments from the original post on X (Twitter)

@AlphaSignalAI
Repo: Check out to get a daily summary of top models, repos, and papers in AI. Read by 280,000+ devs.
2 likes

@Pritesh
This is a really good callout. Very easy to add to any document upload pipeline and eliminate risk early on!
1 like

@homeMetaX
This is basically a security mindset upgrade encoded into software: don’t trust labels, trust evidence. By reading raw file content, the system removes one of the oldest and simplest attack vectors in malware delivery
1 like

@PromptSlinger
1MB is wild for something doing content analysis. most antivirus heuristics are bigger than that and way dumber