WorldFacebook has tolerated hate speech in countries at war for its growth

Facebook has tolerated hate speech in countries at war for its growth


Frances Haugen on Monday in London, while testifying to a committee of the British Parliament.DPA via Europa Press (Europa Press)

Facebook became a sorcerer’s apprentice whose success got out of hand. The disclosure of internal documents of the social network according to which its executives allowed, by action or omission, the publication of disinformation and controversial content turned like a boomerang against the company. This information provoked the worst reputation crisis in its history, which already included obscure episodes about data privacy, such as the case of Cambridge Analytica.

Like tobacco companies in another era, or like the pharmaceutical company that manufactured the drug responsible for the serious opioid crisis in the US — all denying the possibility of addiction —, Facebook faces the great moment of truth: after the idyllic global community that intended to create, he contributed to harming coexistence by ignoring potentially violent content, such as hate speech in countries where these messages can have serious consequences, as indicated in the revealed documents.

The finding of negligence doesn’t just apply to Donald Trump’s USA — with the height of fake news thanks to social media, and consequences like the attack on Capitol Hill on Jan. 6—or the still small percentage of vaccinations against covid-19 because of unscientific theories that Facebook helped foster. The company’s lack of control over its content has also wreaked havoc in India by leveraging the new policy. hinduization of the country by the nationalist Narendra Modi. And in Myanmar, stirring up persecution against the Rohingya community. And also in Afghanistan, Yemen, Ethiopia.

This can be seen from the latest set of internal documents leaked to a consortium of international media outlets, the so-called Facebook Papers. A key issue is to determine whether the very functioning of the platform gave rise to the problem; that is, if the tools that made Facebook what it is today — the like and share content buttons, so intuitive — have exponentially multiplied the risk in the absence of adequate moderation of posts.

The revelations, based on interviews with former employees and internal reports, show how the company’s global expansion — present in more than 190 countries and more than 160 languages, with more than 2.8 billion users per month — has neglected control of the content because, in many cases, an insufficient number of moderators with adequate knowledge of local languages ​​and contexts to identify potentially dangerous—or at least questionable—publications in many developing countries. In 2019, in addition, the company reduced its budget for hiring moderators, in favor of the machines.

The leaks also reveal that the artificial intelligence (AI) systems used by Facebook to prevent this type of content are often ineffective, as are the tools for possible warning of a user. The description shows a hypothetically shielded Facebook against complaints, to which is added, in addition, the carte blanche granted to about five million profiles, considered VIP users, for which there would simply be no moderation rules, as he said last month O The Wall Street Journal. Among the important users would be known representatives of the alt-right American and Trump’s inner circle, like his former strategist Steve Bannon, as well as bloggers and portals informative in its orbit.

The hypothetical free rein given by Facebook —almost always in favor of conservative politicians and positions, as the cases of the US and India demonstrate— takes on a more dangerous character in countries where the risk of instability and violence is real. In an analysis published last year on the internal discussion forum on methods for identifying excesses, one official reported “significant loopholes” especially in Myanmar —Facebook’s role in spreading the hate speech that fueled the Rohingya genocide has been demonstrated as early as 2018— and in Ethiopia, where one of its regions, Tigray, is experiencing an increasingly bloody civil conflict. The sorting done by “sorter” algorithms, which detect inappropriate content, proved useless in scrutinizing messages in the languages ​​or dialects spoken in former Burma and Ethiopia; in this case, with abundant death threats.

Facebook’s security breaches can be summarized in four: the linguistic inability to understand, and consequently to moderate, millions of user posts in non-English-speaking countries; the misunderstanding of their own algorithms; inaction when intervening where artificial intelligence programs do not reach (according to a March report, the company only takes action in 3% to 5% of hate speech cases, and in 0.6% of violent content publications ); and an evident neglect on the eve of the attack on the Capitol; in fact, Facebook deactivated certain emergency safeguards imposed for the November 2020 US elections, and had to quickly reactivate some of them when violence erupted on January 6th. The inability to handle the activity online of the trumpist hordes caused discomfort within the company.

The leaked documents are part of the denunciation of Frances Haugen, a former Facebook executive, a technology giant that most quickly surpassed —in just 17 years— the one trillion dollar mark (5.58 trillion reais) in market value, according to Bloomberg. Haugen appeared on Monday before a committee of the British Parliament, two weeks after giving testimony in the US Congress. Facebook spokespeople have tried to downplay the blow to its reputation by saying in a statement that the company has never placed profit above people’s safety or well-being. On the contrary, “we invested 13 billion [de dólares, 72,6 bilhões de reais] and we have more than 40,000 employees dedicated to just one thing: ensuring people’s safety on Facebook.”

The budget of large companies of software, hardware and artificial intelligence services could reach $342 billion ($1.9 trillion) this year, according to International Data Corp. These AI spending should exceed $500 billion (BRL 2.79 trillion) in 2024, according to the same source.

The denunciations of Haugen and other former Facebook employees put this tech giant before its demons, after having managed to partially overcome a judicial onslaught for monopolistic practices. As, at another time, the US tobacco companies and the drugmaker responsible for the opioid crisis for its aggressive campaigns of marketing —and by the false claim that the drug was not addictive—, Facebook seems, according to the leaked information, to have prioritized monetization without providing it with safeguards. He seems to have neglected the safety, even the physical integrity of many people, in order to earn money. According to the former director for the Middle East and North Africa, the goals of global growth were “colonial”, in the sense of favoring, at any cost, hegemony and domination over millions of digital subjects. More than 90% of active Facebook users live outside the US and Canada.

sign up on here to receive the daily newsletter of EL PAÍS Brasil: reports, analyses, exclusive interviews and the main information of the day in your e-mail, from Monday to Friday. sign up also to receive our weekly newsletter on Saturdays, with highlights of coverage for the week.