Christopher Bouzy, engineer and creator of BotSentinel, explains how inauthentic profiles in extremist campaigns divert the debate on social networks

The engineer Christopher Bouzy is not the son of Sarah Connor (from ‘Terminator’), but he is currently one of the most famous bot hunters in the world. He created the service BotSentinel, where it is possible to check whether Twitter accounts are inauthentic – including automated accounts, fake accounts and more –, and which topics they are driving around the world.

Few countries demand as much from the tool as Brazil. Since 2020, when the fake news inquiry led to some imprisonments, Bouzy was surprised with the volume of bot activity in the Brazilian political debate, which intensified during the Covid crisis. In April that year, he published a list of thousands of accounts that joined the ranks of those following President Jair Bolsonaro as soon as they were created.

He explains that the goal of bots is to capture the attention of people in polarized debates, making it seem like some opinions are more predominant than they actually are, inflating hashtags and driving the debate in the media and among real people.

In this interview, Bouzy explains the bot strategy.

PlenaMata: What is the role of bots in the digital debate?

Christopher Bouzy: It all depends on which goal they are trying to achieve. If they wish to convey a message, for instance, that cutting down trees is not an issue, or that a particular leader is great, then they act in a particular way. If they wish to attack people, then they either make direct offenses or spread personal disinformation. It all depends on the type of message.

A lot of the time, they try to capture the attention and surf the wave of engagement reached by larger accounts. It all depends on understanding whether they are attacking or seeking alignment. They could be trying to establish contact with other large accounts to see if they will expand the message they are pushing. There are times when a large account, of an influencer, will expand political bots. A lot of the time, they know they are doing it and are happy to do so, in an arrangement that is good for both sides. On the other hand, there are bots that attack influencers and journalists, seeking to silence them or trying to bring down their posts through reporting to the platform.

How much of the digital engagement in extremist campaigns depends on bots?

They are extremely active on social media when it comes to Brazilian politics, Covid and climate change. Their activity level is extremely high. A few days ago, as the campaign for the second round in Brazil intensified, we saw a lot of activity of fake accounts spreading disinformation.

Who operates these accounts?

We try not to get into the issue of attribution. We know there are public relations companies in the United States, in Brazil and abroad using this type of service, but we only look for the source when that is very important to what we are studying. In any case, we know that different groups have different ‘signatures’. Some of them use avatar generated by artificial intelligence. Others push a given hashtag that only that particular group uses. We then identify the accounts belonging to that group.

Is there something specific about the bots acting to support Bolsonaro compared to the other patterns observed by BotSentinel around the world?

We cannot say bots respond specifically to the campaign, because bots work in ways that can be very different. Well, the most specific thing about them is that they tweet in Portuguese. But there are many accounts posting both in Portuguese and in English. We believe they sometimes translate the same message into English in order to expand their message globally.

Bots are not always homemade, the agencies can hire them abroad. Many companies work for more than one country at the same time.

In April, you published a long list of recently-created accounts that immediately joined the ranks of those supporting Bolsonaro. In the electoral period, there have also been many recently-created accounts following people. What is the role of these new accounts?

They could be replacing suspended accounts. In other cases, the idea could also be to flood the social media with accounts supporting Bolsonaro, Trump of whoever it is. This happens a lot when an election is approaching. They follow specific individuals or push specific messages.

For those who are not using a service such as BotSentinel, what are the best indications that an account could be a bot?

One of the main ones is to check whether the account is too dedicated to only one type of message. A real person posts about politics but also talks about other things, like sports, films, their family. If the account is pushing the same messages and engaging with the same accounts all the time, it is very likely that it is inauthentic. And inauthentic is different from automated, because it could be a character created by an agency and operated by someone. When we think about the term “bot”, we think about something completely automated, but it is not like that. A lot of the time, the accounts are controlled by human beings.

How can we revert this situation?

It is unfortunate that our reality has reached this point. It is bad for democracy, because these accounts can influence people. We already see this happening every day. It is necessary to pass better, more efficient laws to deal with this issue. Twitter itself has been very proactive in taking down the more obvious cases compared to other platforms, but there are many profiles flying under the radar. I believe that the platforms, particularly Twitter, need to be more proactive.

The Digital Democracy Room at FGV ECMI is an initiative to monitor and analyze the public debate on the internet. Currently, it has partnerships to help monitor politics on the networks in Brazil and in Latin America. This content was produced by the partner InfoAmazônia.