When it comes to online disinformation and threats to democracy in 2024, we tend to focus on fake news from AI. But a new New York University report examining the digital landscape finds that the greatest danger remains a more familiar and persistent one: social media.
Three years after the January 6 insurrection in Washington and a year after the January 8 violence in Brazil, the companies behind major web platforms, such as Meta, Google and contributed to these events, according to the authors of the report from the Stern Center for Business and Human Rights at New York University.
The report concludes that there is a “step backwards”.
Based on an analysis of academic literature, relevant news reports, and interviews with a dozen current and former executives of social media companies, the report aims to catalog the different ways in which online tools and platforms could be used to subvert the 2024 US elections and other upcoming elections around the world. He points out hyperfakes (deepfakes) generated by AI as an increasingly problematic vector of propaganda and deception. But “there is a more fundamental problem,” said its lead author, Paul Barrett. Social networks are still designed to disseminate information of all kinds quickly and widely, and there are indications that they are backing away from their commitments to preserve public discourse by eliminating misinformation, when they should be doubling down. efforts in this direction.
The most egregious example is at X, formerly Twitter, where Elon Musk cut staff, dismantled the election integrity team, and reinstated accounts previously banned for hate speech and spreading baseless claims. At the same time, Elon Musk helped fuel suspicion of content moderation more generally, portraying efforts by Twitter and other tech companies to detect and remove election-related falsehoods as part of a left-wing effort to suppress free speech. X did not respond to our request for comment for this article.
Mr. Musk’s actions, combined with widespread layoffs across the industry and Republican-led hearings and lawsuits aimed at blocking coordination between social media companies and the government, have contributed to lower standards for rival social media platforms, Barrett says.
Meta, which owns Facebook, Instagram and now Threads, maintains a large, industry-standard content moderation workforce and continues to enforce policies regarding election interference. But the New York University report notes that the company also reinstated Donald Trump’s account after banning him following the events of January 6, 2021, and relaxed its rules to allow political ads to be put back. questioned the legitimacy of the 2020 election. He also cites press reports from 2022 that the company’s election team was reduced from 300 to 60 employees during the wave of layoffs totaling 21,000 employees. Since then, the CEO, Mark Zuckerberg, has reportedly stopped meeting her regularly.
Meta’s Andy Stone disputed that statistic, and the company has since detailed how it plans for the 2024 U.S. election. Meta spokesperson Corey Chambliss noted that Trump’s reinstatement also comes with new restrictions and noted that the company estimates that less than 3% of Facebook users’ News Feed content is political in nature.
YouTube also reinstated Trump and reversed its policy of removing videos questioning the 2020 results, although it still demonetized them. Finally, there is TikTok, which the report identifies as an imponderable factor in the 2024 election. While it is clear that the platform has become more influential in politics since 2020, Mr. Barrett recalled, we do not know exactly how this influence will manifest itself in practice. While “their policies are all good,” Mr. Barrett said, there are concerns that video disinformation is more difficult to moderate at scale than text or images, and questions remain about links of the application with its Chinese parent company and the Chinese government.
According to Mr. Barrett, the big platforms seem to have rallied around a kind of “collective fatalism” about their inability to appease critics, no matter what they do.
“If they fully adopt this attitude, we will be in big trouble,” he added. Indeed, some of these problems are inherent in the way they have built their systems, such as algorithms that cause sensationalist content to go viral regardless of whether it is true or not, and outsourcing of moderation. contents. “And if they stop trying to fix it, that’s a perfect recipe for the same problems to persist.” »
The report’s recommendations will come as no surprise to those who have heard social media critiques over the past five years. These recommendations include hiring and onboarding content moderators, funding more external fact-checkers, installing “kill switches” for the most viral content, and “preparing for a crisis” like that of January 6, 2021. Mr. Barrett acknowledged that businesses may not be inclined to respond to these calls in the current climate, but he argued that pressure must still be kept up .
The report also contains recommendations for the U.S. government, including strengthening protections for election workers, who increasingly face harassment and violent threats, both online and offline.
Anika Collier Navaroli, a former Twitter policy chief turned whistleblower who is now a senior fellow at Columbia University’s Tow Center for Digital Journalism, said she hopes tech companies will heed the call to prepare for a repeat of the events of January 6, or even worse. While she hoped her testimony before the commission on the assault on the Capitol would spark reflection on the role of social media in the insurrection, the industry today is “still in pre-2020 conditions.”
“All the ingredients are there, everything points” to a new attempt to discredit the election and foment political violence, she said.
Mme Navaroli explained that one of his “nightmare scenarios” is a “November surprise” in which a foreign influence operation uses an AI-generated fake to deceive people just before the US presidential election. But another possible scenario is simply a repeat of the events of January 6, 2021 and January 8 that would take advantage of the platforms’ inability to meaningfully adapt.
“I don’t want to be right,” she confided. I hope someone reacts. Let’s do something to prove me wrong. »
