Algorithmic racism: Social media boosted influencers reach after posting photos of white people

Editor's note: Originally published in 2020, this story was among the first reports in a major Brazilian news outlet to examine the concept of algorithmic racism and its impact on digital platforms. At the time of publication, R7 was Brazil's third most-read digital news outlet, helping introduce the debate on algorithmic bias and racial inequality to a broad national audience.

Beginning with a seemingly ordinary question about social media visibility and creators' earnings, the story gradually expands into a broader examination of algorithmic bias. It concludes by asking what racial inequality looks like when public security decisions are increasingly shaped by AI-powered technologies.

Healthy lifestyle, yoga poses, a beach house, children and thousands of likes. If that description made you picture a white woman, it is not because there are no influencers from other racial backgrounds who fit the same profile.

Sá Ollebar, a Black content creator, is one such example. Last week, she said her Instagram reach increased by 6,000% after she posted photos featuring white people on her profile.

"I used to think my posts weren't reaching people because I didn't have the perfect house, the perfect forest or live by the beach," she wrote on social media. "Now I live by the beach, I've built a nearly 400-square-meter house, I have more than 600 plants—and guess what? It still doesn't matter."

For those unfamiliar with the term algorithmic racism, Ollebar's account may sound exaggerated. But according to Tarcizio Silva, author of Communities, Algorithms and Digital Activism: Afrodiasporic Perspectives, her experience illustrates the imbalance of power between platforms and users.

"Every day, hundreds of millions of people create content and generate value—either for free or under precarious conditions—for platforms such as Facebook, YouTube and Instagram, yet receive virtually no meaningful information about how content recommendation systems or visibility rankings actually work," said Silva, a digital media researcher and PhD candidate in Human and Social Sciences at the Federal University of ABC.

In his research, Silva documents cases illustrating how racial inequality manifests across digital platforms.

Examples range from beauty apps that automatically lighten users' skin tones to Google searches for the phrase "Black woman teaching" returning pornographic content. Instagram, meanwhile, relies on machine learning to identify users' preferences and determine which content they see.

"This system reproduces and intensifies the worst features of mass media while masking them behind the appearance of a social network and the opacity of algorithmic decision-making," Silva said.

The color of influence

Research shows white influencers earn, on average, 51.1% more than Black creators.

Just as consumer behavior shaped the advertising industry's hiring of Black models for decades, similar dynamics persist in the influencer economy.

A survey conducted by the agency Black Influence with 760 content creators found that white influencers earn, on average, 51.1% more than Black creators. Audience reach remains one of the main factors brands consider when selecting influencers for commercial partnerships.

"This is fundamentally a social issue," said Ricardo Silvestre, founder of Black Influence and coordinator of the study. "We know Black people have historically earned less than white people, and that inequality is reflected in the influencer market. In many cases, Black influencers are actually more qualified and have stronger connections with their communities."

To illustrate how racial barriers shape success in the influencer industry, Silvestre points to the launch of Rihanna's Fenty Beauty in Brazil. Although the cosmetics brand was designed to serve a wide range of darker skin tones, it faced criticism for promoting its Brazilian launch primarily through white models and influencers.

"Many of the women who had already been using Fenty were eagerly waiting for the brand to arrive in Brazil. But when it finally launched in a country with a majority Black population, the campaign lost its meaning because of a flawed marketing strategy—a symptom of an advertising industry that remains overwhelmingly white." From algorithms to real life

While algorithmic racism directly affects the income and visibility of Black creators, its consequences can be far more severe. Silva points to the growing use of facial recognition technology in policing as one example.

"The expansion of incarceration through increasingly sophisticated surveillance technologies is already a problem in itself," Silva said. "It becomes even more serious because these systems are often inaccurate."

According to Tarcizio Silva, research from the United States, the United Kingdom, Brazil and other countries has documented repeated failures of facial recognition systems, including false identifications, wrongful arrests and the disproportionate targeting of Black and other marginalized communities, increasing their exposure to police violence.

"Brazilian authorities continue awarding questionable public contracts for these technologies. More governments are likely to adopt bans or long-term moratoriums in the coming years, but until then, a great deal of state violence will continue to be normalized."