“It’s even worse than it looks.” This is the answer I’m giving these days to friends who have asked me about the latest Retina Paper “Fake news: a matter of robots”.

Before continuing, I must clarify that there are all kinds of false accounts (also known as bots): good, regular and bad. The same as with Gremlins.  Useful bots are dedicated to gather information from a variety of media sources to create a news feed. Some companies use them to help answer questions from clients.  We are used to talking with these machines. “Press one if you want to make an appointment.” “I don’t think I heard you correctly…”

But what bad robot accounts do is something that we are not yet very conscious of, basically because they have learned to behave like humans. They interact with us as if they were one of us, but we continue talking with a machine. And that machine is programmed with a clear objective: to manipulate us. I think we can all imagine why: to sell more, alter the price of something, influence us politically, steal personal data, sow panic…

How do they do that?

Let’s start from the fact that in social networks we are not able to distinguish a robot from a human. Obviously, that does not happen to us in the real world. In the virtual world, people tend to trust us more in general (especially if that person is “popular”, even if its thousands of followers are also fake.)

If you were walking down the street and a random guy tells you that there is life on Mars, you wouldn’t believe it, right? What if you read on Facebook or Twitter that a person with many followers is saying that there is life on Mars? Is it possible that you value the veracity of that information in another way? It is a somewhat extreme example, I admit, but imagine that they say something like: Fulanito has died.  Or, there is a terrorist attack on the White House.  Or, they have detected cases of birth defects from the “x” vaccine.

Bad bots are very smart when it comes to grabbing us with fake news because one of the tactics they use is to mention accounts of famous people. “There is life on Mars @nytimes” seems like a response to a piece of news published by the New York Times.

The robots and those who program them are even smarter than that. They know by our profiles what we like, what we think about various issues.  Then they bombard people who are most likely to believe and spread the fake news.  That is, they attack the most vulnerable in each case to spread their virus.

Bots have played an important role in many instances.  Has anyone heard the story of a mini-crash that the New York Stock Exchange experienced because of automatic operations? It was the largest in its history, but it lasted a short time.  An error in bots programmed to operate in the stock markets created the panic in a “jiffy”. It is not very clear why they suddenly went crazy.

The fact is that there are virtual stockbrokers who try to find useful information about buying and selling shares by tracking social networks and the internet.  If a large group of important people (measured by number of followers) start talking on Twitter about a specific product or service, these vigilant robots are able to detect the trend and start taking positions in the markets. Something like this happened with a tech company called called Cynk. Many people started talking about it on the internet and social networks and the bots noticed. They started buying shares of the company. The prices rose quickly. Cynk’s worth rose to 5 billion in stock. But it was all a lie: the conversation was artificial, it had been fed by well-trained robots. When the trick was revealed, the securities fell and some investors recorded real losses.

Second hand bots

These ghost accounts can be reprogrammed. To put it in very simple terms: they can be activated by detecting the word “Trump” and can be re-programmed to be activated when they read “Macron”. It’s not science fiction. This has already happened. Some of the fake account networks that disseminated news that favored US candidate Donald Trump went to “sleep” after his victory.  But … someone reactivated them to spread news against Emmanuel Macron in the French elections months later. That is, there is a black market for these zombie and bad accounts. This is clear from the research of Emilio Ferrara, a professor at the University of Los Angeles.

What did those bots want? Well, to alter public opinion to what the programmer wants (nobody knows exactly where they are coming from, but many bots point to Russia lately). And if they cannot convince people, at least for a while they have generated enough noise to keep the audience looking elsewhere.

Quick guide to detect them

If at this moment you are scared to death, here is a short list of ways to detect them on Twitter:

  1. Bots retweet more than humans.
  2. Bots have longer names.
  3. Bots tweet and are retweeted less

These are some of the basic characteristics, but researchers do not yet know how to distinguish these robots in social networks.  They began with rudimentary techniques, similar to putting mousetraps in doors. Sure, the most “clueless” bots were detected. But the level of sophistication is very high in other cases. The companies Facebook, Twitter and Instagram, do not want to have ghost accounts and are supposed to fight them at all levels (without being able to clarify their tactics because that would give the enemy an advantage) to stop the propaganda of these robots.

It is possible that in the future the necessary tools will be developed to turn these bots into what is now spam mail: a very smart machine capable of removing those useless and deceptive emails from us. But we are not there yet. And we do not know if we will get there … Until then, dear friends of the virtual world, keep a watchful eye.

Thanks to Dan Thomsen for translating!


Deja un comentario

Tu dirección de correo electrónico no será publicada. Los campos obligatorios están marcados con *