fake news, facebook, fake news elections
Facebook's artificial intelligence experiment was shut down when chatbots made up a language all their own. Photo: Reuters

A review of Facebook's internal practices for policing hate speech on the social network found that they are inconsistent, appearing to favor white men, political elites and governments over minorities and grassroots activists.

 

The investigation, conducted by the independent journalism foundation ProPublica, looked at hundreds of rules the company has established over the last decade to make Facebook a safe place online. Although the platform has had its moments as a vehicle for social change — it's credited with assisting the anti-authoritarian Arab Spring uprising in 2010 — it has drawn increasing scrutiny as a vehicle to spread fake news stories and as a recruitment vehicle for terrorists.

 

The company employees 4,500 "content reviewers," but the guidelines for censoring posts are elaborate and can have questionable results.

 

ProPublica points out:

 

— After the June terror attack in London, U.S. Rep. Clay Higgins (R-LA) wrote a Facebook post calling for the slaughter of “radicalized” Muslims. “Hunt them, identify them, and kill them,” he wrote. “Kill them all. For the sake of all that is good and righteous. Kill them all.” The post was not removed.

— In May, when Black Lives Matter activist Didi Delgado wrote “All white people are racist. Start from this reference point, or you’ve already failed,” the post was removed and her Facebook account was disabled for seven days.

The ProPublica report described a training document for "content reviewers," which instructs them how to apply Facebook’s hate-speech algorithm. The policy is to delete "curses, slurs, calls for violence and attacks only when they are directed at 'protected categories'—based on race, sex, gender identity, religious affiliation, national origin, ethnicity, sexual orientation and serious disability/disease. It gives users broader latitude when users write about “subsets” of protected categories," the report said. "So white men are considered a group because both traits are protected, while female drivers and black children, like radicalized Muslims, are subsets, because one of their characteristics is not protected."

Adding to the complexity: A president who has advocated racist views as political policy. According to the internal documents ProPublica saw, Donald Trump’s vow to ban Muslim immigration violated Facebook's policies against “calls for exclusion” of a protected group. But Trump's posts stayed up, exempted by Facebook founder Mark Zuckerberg to allow for political discourse, the Wall Street Journal reported. 

Last month, Zuckerberg announced that Facebook will nearly double its team of "content reviewers" to 7,500, up from 4,500. "Their work amounts to what may well be the most far-reaching global censorship operation in history," said ProPublica. "It is also the least accountable: Facebook does not publish the rules it uses to determine what content to allow and what to delete."