Video transcript
QUEM MANIPULA AS ELEIÇÕES? | Não dá tempo de explicar
78 statements · 1 politicians · June 18, 2025 · 8 min
Statements by Guilherme Boulos. 78 transcribed statements, with topic and stance on the ones the analysis classified.
Speakers
Click any statement beside the video to jump straight to the moment it was said.
Transcript
78 statementsFull transcript of the video. Statements the analysis classified carry topic and stance.
who really manipulates elections in Brazil and in the world.
No, no, I'm not going to jump on the wave of what we've been hearing in recent years about fraud in the voting machines, the manipulation of the TSE (Superior Electoral Court)'s secret room.
No, it's not fake news, not a conspiracy theory, none of that.
But there is indeed a force that alters and influences elections in democracies around the world.
I'm talking about the big techs, the large technology platforms that run the social networks.
Look.
There was an Oxford study that came out last year examining how the social networks' algorithm operates.
If the social networks' algorithm isn't neutral, I think many attentive people already know that.
Now, the algorithm is designed to promote extremist content, content that generates strong emotions—anger, hatred, fear—polarizing content, for one reason.
Money.
See how it works.
The algorithm is an artificial intelligence that was shaped by machine learning dealing with people's psychology, the profiles of users on social networks.
And it was noticed that stronger content of this nature generates more interest, generates more clicks.
It's the famous clickbait.
From that, the algorithm of Instagram, TikTok, X, or any other, causes content of that nature to be distributed to more people.
Ah, but what does that have to do with money?
Everything.
Because you receiving this content that pushes you to stay on the network more often means you stay longer on the platform, more time on the screen.
And you spending more time on the screen means more profit for the big platforms.
Both because with more users the platforms can absorb more data—your patterns of likes, shares, comments, which is their main form of profit—and because they expose you to more ads, which is a decisive source of revenue for the platforms.
Does that sound a bit abstract?
So let's look at concrete cases of how they acted in recent electoral processes.
The most emblematic case is almost 10 years old: 2016.
I imagine you've heard of the so‑called Cambridge Analytica.
Cambridge Analytica was a company created by Steve Bannon, that guy who gave a Nazi salute and who was one of the main people responsible for Trump's first election.
in 2016.
What did Cambridge Analytica do?
They bought data from Facebook.
At the time Facebook was the main social network in the whole world.
And that segmented data allowed the Trump campaign to deliver a tailored message to each audience.
For example, they would group together a set of profiles based on segmentation—what you like, the videos you watch.
Of people who are afraid, almost terrified of being robbed.
And then they would take that segment—say, I don't know, 10 million Americans have that—
and send those profiles specific pro‑gun content, encouraging the loosening of gun restrictions, linking it to Trump, to address that audience's fear and desire.
Another audience, for example, was having problems with unemployment or low wages.
Then they sent... for that audience, a Trump message saying that immigrants entering the United States were to blame for unemployment.
Thus, they reached 87 million profiles, almost half of the U.S. electorate, sending this kind of content.
Trump won the election.
That same year, something similar—a kind of test before Trump's election—was Brexit in the United Kingdom, where Steve Bannon and Cambridge Analytica operated in the same pattern.
Ah, but that's over, it caused a scandal, a CPI (Parliamentary Inquiry Commission) in the United States, after which the networks stopped selling that data.
AI classification from the transcript of a public video. See the Methodology