Incorrectly identifies some binary files as code or text

[noway]

I've generated 2435 files that are classified as text and code, although under manual inspection the files are binary.

Here is the repo with the generated files: https://github.com/noway/magika-binary-misclassification-as-text

I've used fuzzing. More info here.

Cheers!

[devilkadabra69]

.bin file is not listed in supported files

[noway]

@devilkadabra69 magika should classify those files as Unknown binary data (unknown). It's in the table you've provided, see index 100.

[reyammer]

Yep, random stream of bytes should fall under unknown. Thanks @noway for reporting and for the fuzzing, interesting stuff. Marked this issue with proper tags so that we remember about it when we plan for next iterations. Thanks!