First of all, what is Mágika exactly? Well, it is an innovative AI-based system that the search giant has designed to revolutionize the way binary and textual file types are identified.
Magika stands out for its ability to provide precise file identification in milliseconds, even when running locally on a computer.
Magika employs a customized and highly optimized deep learning model that has been meticulously designed and trained using Keras. This model is remarkably lightweight, weighing only 1 MB.
How Google uses Magika in its day-to-day
The performance of Magika is impressive. When evaluated in a benchmark test of 1 million files covering over 100 file types, Magika outperforms existing tools by approximately 20%.
The system performance is even higher in the case of text files, including code and configuration files, which traditionally have posed problems for other tools.
Internally, Google has already used Magika to enhance user security. The system is deployed at scale to direct Gmail, Drive, and Safe Browsing files to the appropriate security scanners and content policies.
With Magika, Google has observed a 50% improvement in the accuracy of file type identification compared to previous systems that relied on manually crafted rules.
This increase in accuracy has allowed scanning an additional 11% of files with specialized AI malicious document scanners and reducing the number of unidentified files to 3%.
By launching Magika, Google aims to help other software programs improve their accuracy in file identification and provide researchers with a reliable method for identifying file types on a large scale. The code and model for Magika are already freely available on GitHub under the Apache2 license.
