The current meta is Training materials obtained by The Intercept reveal that users have been given new freedom to post a wide range of derogatory comments about race, nationality, ethnic group, sexual orientation, and gender identity.
Examples of new allowed speech on Facebook and Instagram covered in the training materials include:
“Immigrants are dirty, filthy shit.”
“Gay people are weird.”
“Look at that transsexual (below the photo of the 17-year-old girl).”
The changes are part of a broader policy shift that includes suspending the company’s fact-checking program. Mehta said Tuesday that the goal is to “allow more speech by lifting restrictions.”
Joel Kaplan, Meta’s newly appointed head of global policy, said in a statement that the initiative was aimed at “completing complex systems for managing content on the platform, which are becoming increasingly complex to enforce. He explained that it is a means of correcting the situation.
Kaplan and Meta CEO Mark Zuckerberg have touted the change as a way to give users more freedom to engage in ideological dissent and political debate. Previously unreported policy documents reviewed by The Intercept show the extent to which purely insulting and dehumanizing rhetoric is now accepted. .
This document provides an overview of the hate speech policy changes for those working with meta user content and explains how to apply the new rules. The most important changes involve selected “related examples”, i.e. hypothetical posts marked “allowed” or “deleted”.
When asked about the new policy change, Meta spokesperson Corey Chambliss referred The Intercept to comments on Kaplan’s blog post announcing the change. Political discussions and debates occur frequently. It’s not true that we can speak on TV and in Parliament but not on our platforms. ”
“It is a travesty and a lie to pretend that these new rules are more ‘neutral’ than the old ones.”
Kate Kronnick, a content moderation policy expert who spoke to The Intercept, said she framed the new rules as less politicized, given the latitude it gives them to attack conservative bogeymen. I object.
“Drawing the line around content moderation has always been a political undertaking,” said Kronick, an associate professor of law at St. John’s University and a scholar of content moderation policy. “It is a travesty and a lie to pretend that these new rules are more ‘neutral’ than the old ones.”
She said the changes announced by Kaplan, who served as White House deputy chief of staff under George W. “It is a capture, especially since the changes are pandering to a particular political party.” ”
Meta’s public community The standards page says that even under the newly relaxed rules, the company still protects “refugees, immigrants, migrants and asylum seekers from the most serious attacks” and “does not support people of any race, ethnicity or national origin.” It prohibits “direct attacks” against people based on the law.” , disability, religious affiliation, caste, sexual orientation, gender, gender identity, serious illness. However, the helpful examples provided in the internal documentation show various comments that disparage people based on these characteristics marked as “allowed”.
In some cases, the examples provided may seem complicated or contradictory. One page notes that it is still prohibited to make “generalizations” about any group of people, such as “all Syrian refugees are rodents”, using analogies to animals or pathogens. But the comparison to “filth and feces” has now been downgraded from hate speech to a less serious “insult,” which only violates company policy when directed at a protected group. According to an example provided by Meta, this change allows users to say “immigrants are dirty, filthy pieces of shit,” even though “comparisons to acts of subhumanity” are still prohibited elsewhere in the document. It has become possible to broadly dehumanize immigrants with statements such as “You’re a bastard.”
The company’s policy on nausea-based hate walks a particularly fine line. According to the document, “Immigrants make me want to vomit” is allowed, but “Muslims make me want to vomit” should be removed because it claims to be a group. cause illness. ”
Comparisons to common animals are still against the rules, but many other kinds of broad and hateful stereotypes are now allowed. “All behavioral statements (eligible and ineligible)” also no longer violate meta rules, allowing blanket generalizations linking entire races and ethnicities to crime and terrorism, the document says. . Examples of acceptable racial generalizations in this document include, “I can’t trust these damn immigrants, they’re all criminals,” and “I bet Jorge was the one who stole my backpack after track practice today.” etc. are listed. “All immigrants are thieves” and “All Japanese are yakuza.” However, the paper points out that statements such as “all black people are drug dealers” are still prohibited under the new rules.
Other sections of the document list examples of prohibited “insults related to sexual immorality,” such as “Jewish women are sluts.” But the document also provides a wealth of examples of newly permissible insults targeting specific gender identities and sexual orientations, such as “homosexuals are sinners” and “transgender people are immoral.” are. Posts such as “Lesbians are so stupid” will still be banned as a “mental insult”, but posts that say “Transgender people are mentally ill” will still be allowed.
Generalizations about superiority and inferiority are similarly complex, but attacks on immigrants tend to get a pass. Examples of prohibited content include “Christian men are completely useless” and “Is that me?” Or are all autistic women ugly? ” and “Hispanics are as dirty as the ground we walk on.” while “Mexican immigrants are trash!” would now be considered acceptable.
Overall, restrictions on claims of ethnic or religious superiority have been significantly relaxed. The document states that meta currently refers to “a claim of superiority (a) based on innate intellectual ability, and (b) unless substantiated and does not refer to the inferiority of another (protected characteristic) group.” It is explained that “statement” is permitted. Permissible statements under this rule include statements such as “Latinos are the best!” and “Black people are better than everyone else.” “Black people are more violent than white people.” “Mexicans are lazier than Asians.” “Jews are much greedier than Christians.” Comparative claims such as “are now accepted. The example “white people are more intelligent than black people” is off-limits simply because it involves intellectual ability.
However, general statements about intelligence appear to be permissible as long as they are shared with the purported evidence. For example, “I read a statistical study that says Jews are smarter than Christians. As far as I know, that’s true!” Are you required to link to such a study, or just claim that it exists? It is unknown whether
Rules regarding explicit expressions of hatred have also been significantly relaxed. “Disparaging, hateful, or dismissive statements such as ‘I don’t like’, ‘I don’t care’, or ‘I don’t like’ are now considered nonviolent and are allowed,” the document says. Acceptable examples include posts that say “I don’t care about white people” or “I’m a proud racist.”
The new rules also prohibit “targeted abuse” against protected groups, “including the use of the word ‘shit’ and variations thereof.” For example, a post that says “Oh, the damn Jews are at it again” is against the rules simply because it contains obscenity (the new rules include “bitch” and “motherfucker”). approved for use).
“Referring to targets as genitals or anus is now considered not an offense and is permitted.”
Another change in policy is that “referring to objects as genitals or anus is now considered not a violation and is now permitted.” As an example of what is currently acceptable, Facebook suggests “Italians are crazy.”
Although many of the examples and underlying policies seem confusing, this document provides clarity on the permissibility of derogatory remarks about transgender people, including children. Noting that “‘tranny’ is no longer a designated slur and is no longer a violation,” the material cites three examples of speech that should no longer be removed: “tranny is the problem,” “that tranny Look (below photo) “17 year old girl”, “Kick this transsexual out of school (below photo of high school student)”.
After this article was published, Meta spokesperson Andy Stone told The Intercept that one of the examples in the company’s information materials was incorrect. Stone said the meta meant to explain what content was not allowed when it included the example, “Look at that tranny (below the photo of the 17-year-old girl).”
According to Gillian York, director of international freedom of expression at the Electronic Frontier Foundation, said meta hate speech protections were historically well-intentioned but deeply flawed in practice. Yorke said: “While this often results in the over-moderation that I and many others have criticized, these examples demonstrate that meta policy changes are political in nature and are simply expressive.” “It shows that it is not intended to increase the freedom of people.”
Meta has come under international scrutiny for its efforts on hate speech, particularly the role hate speech and other dehumanizing expressions on Facebook played in fueling genocide in Myanmar. It is attracting attention. Following criticism of Facebook’s mishandling of Myanmar, which the United Nations found to have played a “critical role” in the genocide of more than 650,000 Rohingya Muslims, the company has announced plans to prevent similar rhetoric from spreading in the future. has spent years promoting its investments.
“The reason so many of these lines are drawn in their original place is because hate speech often goes beyond just speech and turns into real-world behavior,” said content moderation scholar Chronick. says.
This is a premise that Mehta claimed to share until this week. “We have a responsibility to combat abuse on Facebook. This is especially true in countries like Myanmar, where many people are using the internet for the first time, and where social media can spread hatred and increase tensions on the ground. could be used for multiple applications,” Sarah Hsu, the company’s product manager, said in a 2018 blog post. “We are changing our approach to fake news to take account of the changing situation, but the rules on hate speech remain the same: it is not allowed.”
Updated: January 9, 2025, 9:11pm ET
This article has been updated to include comment from Meta spokesperson Andy Stone, who notes that a mistake in Meta’s internal documentation led to a post about transgender youth being incorrectly classified as acceptable. Ta.
