A little over a week ago, Google announced a new Google Drive policy. It stated that the tech giant would start to actively search for files breaching its abuse program policies and/or Terms of Service and locking the ones that do.
Simply put, if such a file was found and restricted, the owner wouldn’t be able to send it to others, and all access to the file, except that of the owner, would be revoked. Google stated that this change had the aim to control the spreading of malware, hate speech, explicit content, etc.
That said, complaints about Google Drive locking files that were neither of those things started piling up recently. From the looks of it, it seems that the AI system in charge of sifting through which files should be deemed dangerous is acting up and failing to do its job properly. Algorithms—the ultimate example of a double-edged sword.
There are even examples of .txt files containing a single number inside that are falsely marked as inappropriate. One of the best examples comes from a Michigan State University assistant professor Dr. Emily Dolson.
Dr. Dolson was uploading a large set of files to Google Drive while preparing to send her students’ programming assignments. The files consisted of the “inputs and the expected outputs” of said cases. “Shortly after uploading them,” she explains ”I received a string of emails from Google indicating that those files had been flagged for copyright infringement.”
Now, in its announcement of this new policy, Google has stated that in situations such as the one with Dr. Dolson, the user should “receive an email with details and potential actions they can take to request a review.” Unfortunately, however, that did not happen in her case.
After posting her issue on Twitter, a Google engineer going by Misha Brukman replied to Dolson, assuring her that they are working on a solution.
