Facebook relies heavily on artificial intelligence to detect spam, hate speech and more. It’s a good idea on paper, but it often leads to silly mix-ups that confuse and anger users even more. The latest example occurred this week when Facebook accidentally flagged the U.S. Declaration of Independence as hate speech.

The controversy began when a local newspaper in Liberty, Texas, called The Vindicator decided to post the Declaration of Independence in small parts leading up the Fourth of July. However, Facebook took down part of the Declaration that referred to “Indian Savages.” Facebook eventually apologized and reinstated the post, but not for a while — and not before confusing the paper’s editors.

“If any human being working at Facebook were to review it, no doubt the post would be allowed,” Vindicator managing editor Casey Stinnett wrote. “The editor has searched for a means of contacting Facebook for an explanation or an opportunity to appeal the post’s removal, but it does not appear the folks at Facebook want anyone contacting them. Or, at least, they do not make it easy.”

With all of these censorship issues, Facebook’s greatest problem is a lack of clarity with its decision-making. Not to mention the fact that the tech giant holds all the power, and it could seriously hurt the business of any publisher using its platform. Here’s hoping the humans at Facebook can step in to stop snafus like this in the future.