Recently, Google officially announced the release of the stable version 1.0 of its AI-based file type detection tool, Magika. The latest iteration of this open-source project introduces significant core architectural upgrades aimed at improving speed, security, and accuracy.

QQ20251107-144122.png

According to reports, Magika's core advantage lies in its use of a highly optimized custom deep learning model to accurately identify file types, solving the common misclassification issues faced by traditional file signatures or heuristic methods. Google has used Magika extensively internally, integrating it into core services such as Gmail, Google Drive, and safe browsing to enhance security checks during file uploads. The most significant update in version 1.0 is the complete rewrite of its core engine using the Rust programming language.

QQ20251107-144154.png

This move significantly improved the tool's operational efficiency and memory safety, ensuring that even on standard CPU environments, precise file identification can be completed in just a few milliseconds. In terms of accuracy, Magika performs exceptionally well, with an overall accuracy improvement of about 30% compared to traditional methods. Particularly in identifying potentially malicious content such as VBA macros, JavaScript, and PowerShell scripts, the accuracy reaches an impressive 95%, providing strong support for cybersecurity defenses.

https://opensource.googleblog.com/2025/11/announcing-magika-10-now-faster-smarter.html