A couple of weeks ago, X – a Hungarian user living in Berlin – was browsing through Tumblr, a microblog website, on his mobile phone when he stumbled upon a funny video. Though its content is actually completely irrelevant to this story, just to set the record straight, the video contained a human arse – most likely a woman's – with some make-up resembling a human face, moving in a way to make it look like it was eating some candy being dropped on it. Those interested can watch the post here. No worries, no human genitalia visible. I'll just include the video here:

link Forrás

X found this rather funny and – using the share icon of the Tumblr-application for smartphones – sent it to one of his friends on Messenger. (Messenger is owned by Facebook).

Next day, when X woke up and checked his Facebook, he found a message notifying him that his account had been blocked.

He couldn't post, like or browse and was even banned from sending private messages on Messenger.

The message contained the following:

He checked the given link through which he learnt that he had violated Facebook policy regarding nudity. The message was clearly phrased as if X had publicly shared a prohibited human body part in a public Facebook-post. X was confused since his timeline is automatically set to private and he did not post anything controversial, the only possible explanation was the link which he had sent via chat.

This is the point when he contacted me. X already had an idea - I did not - that Facebook is not just capable of viewing people's private messages but actually evaluates them based on the content and censors what it doesn't approve of and even goes onto to punish users. X had a theory that this is probably an error in programming.

After I had heard his story, I contacted the European regional representatives of Facebook who promised a prompt investigation, then I didn't hear from them for days. Since the whole case seemed to be stuck, a few days later, X ran his own test which could rule out the possibility of him being practically unconscious and accidentally posting the video with the arse on his wall, perhaps resulting in the ban. X knew nothing like that had happened but the whole thing seemed so terrifying, Facebook monitoring everyone's private conversations, that he decided to test himself first.

The test was basically this: he sent the very same link via Messenger to the same friend. There could not have been any mistakes involved this time. 20 minutes later, a notification of a 3-day ban arrived as well as two other messages which referred to nudity on his timeline, similarly to the first incident.

Days later, X. received a letter from a Facebook employee (one I had also contacted earlier and whom we set up a meeting with immediately after the first incident.) According to the letter, the internal investigation conducted by Facebook concluded that X posted nude content. However, X, already knew that this cannot be the case. After he and I – in separate messages – asked Facebook about the inconsistency of their explanation and mentioning the user's private test – said Facebook-employee was attentive at first, and promised further examination of the issue, seemed to be enthusiastic, but after a while, we did not receive any more responses from him.

Both X and I reached the same conclusion: Facebook is certainly monitoring and evaluating its users' private messages.

Based on what?

Everyone who has tried to send a porn link via Messenger is aware that Facebook does monitor chat messages. Still, we only had knowledge of a list of banned sites, for example porn websites, ensuring that if one tries to send such content - for example the link to Pornhub - it is automatically banned. I tried this and it indeed worked.

However, the link mentioned in our story pointed to a Tumblr blog, not included on the black list. I sent it to a colleague of mine too and the message arrived. It is possible though, that I'll be banned within a few minutes or hours.

So, apparently Facebook robots monitor all links sent via chat, they open them and evaluate them based on their content.

Otherwise, it would have been impossible to know that the link sent by X contained a human arse. It is easier to imagine that X's 24-hour and later, 3-day long ban was probably the cause of some programme error. It really does seem like Facebook noticed unauthorized content, not just in a message of X, but sensing it on his timeline. Nevertheless, it remains clear that links sent are automatically checked, but in this case, one may have been labeled as a public post erroneously.

This post was originally written in Hungarian by László Szily. It was translated by Dániel Baranya.