In the race to be Brazil’s next president, the final two candidates are appealing to voters not only at rallies but online as well.

On Sunday, incumbent President Luiz Inacio Lula da Silva will face right-wing Senator Flavio Bolsonaro in the final round of the election.

The race is expected to be tight. But in its final days, questions have festered over whether the digital arena has been a level playing field for the two candidates.

The social media giant Meta, which owns Facebook, WhatsApp and Instagram, has faced accusations before Brazil’s top election authority, the Superior Electoral Court (TSE), that it suppressed Lula’s content and favoured material promoting Bolsonaro.

Meta’s products rank among the most popular social media platforms in Brazil, and political ad spending on its sites has grown at a breakneck pace.

An analysis from the Brazilian news outlet G1 found that Meta received at least 288m Brazilian reais ($57m) in direct campaign spending in 2026, double what it received during the 2022 presidential race.

That increase marks a shift in how Brazil runs elections. More than 76 percent of Brazilian voters now get their news online, said Andre K Rodarte, a researcher at the Digital Democracy Centre at the University of Southern Denmark.

With social media eclipsing traditional advertising, Rodarte explained that a growing class of politicians is using provocative, viral messaging to connect with voters.

He pointed to the success of online influencers like right-wing personality Nikolas Ferreira and left-wing activist Erika Hilton, who each were reelected this year to Brazil’s Chamber of Deputies with record numbers of votes.

“You are seeing people transitioning from information-producer roles to politics, and they are just getting more and more power, more and more votes,” Rodarte told Al Jazeera. “If this is the media landscape they are dealing with, it’s unlikely they will adopt a different campaign style, or different style of politics.”

But critics warn that this shift to online campaigning has put political hopefuls at the mercy of Big Tech corporations and their biases.

Effort to regulate

In Brazil, the reliance on online campaigning first took off in 2018, with the election of President Jair Bolsonaro, according to Leticia Cesarino, an anthropologist at the Federal University of Santa Catarina who studies far-right politics online.

Jair, a far-right leader, is the father of this year’s presidential frontrunner, Flavio Bolsonaro.

When Jair first ran for president eight years ago, his platform promised tough-on-crime tactics and a return to traditional Christian family values, which he spread through bite-sized social media posts.

At his 2019 inauguration, supporters chanted “Facebook!” and “WhatsApp!” in unison to show their support for his online presence.

“Since then, the left and pro-democracy actors have slowly come to recognise the importance of such media for elections and learned how to deploy it strategically,” Cesarino told Al Jazeera.

“They still lag significantly behind the far right when it comes to digital communication, though, especially since the kind of logic that underlies the attention economy is structurally beneficial to the kind of alarmistic, simplistic, affect-laden far-right discourse.”

But during his presidential term, Jair used his social media platforms to cast doubt on the accuracy of Brazil’s election system.

After losing the 2022 election, Jair continued to spread false claims that the vote had been rigged. His supporters ultimately stormed three buildings – Congress, the Supreme Court and the presidential palace – in the capital, Brasilia, on January 8, 2023, after Lula was sworn in.

In the aftermath, the TSE ruled that Jair had abused his position to undermine the election and prohibited him from holding public office for eight years.

He is currently serving a separate 27-year sentence for plotting a coup. He is also barred from using social media.

The scandal “really shocked a lot of policymakers and political elites to act”, said Vicky Wyatt, a campaign director at Eko, a coalition advocating for more accountability from tech companies.

Although Brazil had already passed some data privacy and data protection laws, leaders added a slew of new regulations after Jair’s presidency.

Brazilian law now bans depictions of electoral candidates generated by artificial intelligence (AI), and it also requires any political ads using AI to carry a disclaimer. Content that advocates violence, hate speech or gender-based discrimination is also considered illegal.

Only those with official campaign backing are supposed to be able to run ads on platforms like Instagram, Facebook and WhatsApp.

The country, said Wyatt, now has “some of the strictest laws in the world when it comes to guardrails around the use of AI on tech platforms”.

But those laws have been the source of pushback. In a statement to Al Jazeera, a Meta spokesperson framed the restrictions as a potential threat to free speech.

“Meta has long warned that Brazil’s strict digital-content rules increase the risk of over-enforcement and threaten free expression,” the company said.

Tech companies like Meta are supposed to walk regulators through their plans to comply with Brazil’s laws. However, evidence suggests Meta has not met its obligations.

Daniela da Silva, executive director of CTRL+Z, a nonprofit digital rights group, said some campaign coordinators have found their posts are not being shared as widely as they expected.

“They find their organic reach has decreased since the beginning of the elections, and they need to boost their posts,” da Silva told Al Jazeera. She added that some coordinators reported spending more than 70 percent of their budgets for such boosts on Meta.

As of October, CTRL+Z says it has collected 1,100 reports this year of accounts being taken down or having their reach restricted.

It determined that “the cases are mainly related to profiles supporting Lula”, who saw his own advertising account briefly suspended last month.

Lula’s campaign has cited studies by Eko, CTRL+Z and other groups to petition the TSE to intervene and enforce rules that require equal treatment for political campaigns.

Watchdog groups have also questioned whether social media platforms are complying with regulations that require them to remove false information and restrict AI-generated content.

One test, conducted by Eko, asked Meta to approve 15 election-season ads that included AI-generated content promoting violence and disinformation. In 13 cases, the social media platform approved them.

In another study, NetLab, a research group at the Federal University of Rio de Janeiro, found that Meta had allowed more than 1,600 political ads to run on its services, even though they were paid for by companies outside the country.

The group concluded that malicious actors appeared able to submit ads and tag them as non-political, allowing them to run on Meta’s services with little scrutiny, said Debora Gomes Salles, NetLab’s research head.

Meta made 2 million Brazilian reais ($400,000) from those ads, which, as far as Salles could tell, were meant to sow disinformation in the country.

Some told voters, for instance, that they could cast ballots online through a website that collected their private data. Voting online is not allowed at all in Brazil, Salles pointed out.

“Meta wasn’t really able to prevent these operations from happening, and even when it did do something and moderated this content, these ads were still a source of revenue for the company, and they already reached millions of Brazilians,” Salles said. “The damage was already done.”

Meta has repeatedly denied favouring any of this year’s presidential candidates on its platforms.

In a statement to Al Jazeera, a spokesperson said that technical errors were to blame for incidents like Lula’s campaign ads being taken offline temporarily.

The company also disputed the Eko report that suggested shortcomings in how it moderates campaign advertisements. The spokesperson said Meta’s system had correctly identified “content about social issues, elections or politics”.

“We removed ads that violated our policies,” the spokesperson said. The report’s small sample size, “out of millions in our Ad Library, does not reflect the scale of this work”.

Cesarino, the anthropologist who studies the far right online, said digital platforms will likely continue to spread problematic content.

The popularity of such content, she added, has contributed to the rise of far-right figures across the globe.

“It is not by chance that other far-right ‘populists’ rose at around the same time as Jair Bolsonaro in Brazil, such as Donald Trump in the US or Narendra Modi in India,” Cesarino told Al Jazeera.