Meta's new content moderation policy, already in effect in Brazil, allows content considered illegal under the country's law, such as expressions of racial supremacy and religious prejudice.
Expressions like “white people are the best” and “I hate black people” are among those now allowed to circulate on Instagram and Facebook.
Announced by Mark Zuckerberg on the 7th, the shift did not affect just the treatment of gender, sexual orientation and immigration issues — topics dear to Donald Trump's followers and which were emphasized in Meta's announcement.
Some of the changes are in the public version of Meta's policy, which replaced the expression “hate speech” with “hateful conduct.”
However, internal Meta documents to which Aos Fatos had access are more explicit — authorizing posts that reinforce racial, ethnic, religious stereotypes and prejudices and those against people with disabilities, among other groups.
Classified by Meta as “protected characteristics” (PCs), these identity categories have stricter moderation rules (read below).
The application of the new hate conduct policy in Brazil was confirmed to the government by Meta on January 13. The Attorney General's Office considered that the change disrespects Brazilian legislation and called a public hearing for this Wednesday (22).
When asked by Aos Fatos about the guidance to moderators not to block illegal and biased content, Meta just shared a link in which it discloses its new hate conduct policy.
Hatred and disgust
The earliest version of Meta's policy explicitly prohibited the use of hate speech and rejection of protected classes of people, such as:
- “I despise”
- “I hate”
- “I can’t stand”
- “I don’t respect”
- "I don't like"
- "I don't care"
This section was removed from the new text, which now only indicates the prohibition of sentences that indicate disgust or repulsion, such as those containing the construction “it makes me want to vomit.”
Meta's internal documents confirm that the omission reflects a concrete change in the company’s policy, guiding moderators not to remove hateful statements against minorities.
An experiment carried out by Aos Fatos confirmed the new guidelines. On the 15th, the report presented complaints from five different profiles to Facebook against a post that said, “I hate black people” — published by a neo-Nazi page to, supposedly, test the new rules. On the afternoon of last Monday (20), the publication was still online.
The company's rules also began to authorize expressions of superiority and comparisons between protected classes of people, facilitating the speech of white supremacists.
The guidelines for moderators say that statements of superiority and comparisons are permitted as long as they do not refer to the intellectual capacity of one class in relation to another. The caveat, however, includes one exception: if the user presents “justification.”
By rule, although the phrase “whites are more intelligent than blacks” is prohibited, the company authorizes variations citing supposed studies that would prove the statement.
Claims that some races are more intelligent than others are not scientifically founded and have historically been used to justify racism, colonialism, slavery, and racial eugenics.
In the case of comparisons that do not involve intellectual capacities, Meta's rules waive the need to cite sources, allowing the dissemination of biased phrases such as “black women are more aggressive than white women.”
Expansion of attacks
Meta’s moderation guidelines ban dehumanizing speech, such as attacks comparing members of minorities to animals, diseases, and demons. General rules also prohibit allegations of sexual immorality and the linking of certain populations to violent crimes
— such as saying that gay people are pedophiles, for example.
In the case of immigrants and refugees, the rules grant only partial protection, allowing these groups to be the target of insults that are forbidden to others, such as attacks on their character, allegations of uselessness and links to criminal acts of lesser severity, such as robbery.
With the change, the company expanded the list of exceptions to cover also issues of gender and sexual orientation. The following are now allowed:
- Advocating gender-based exclusion for military, police and teaching jobs, such as stating that women should not be accepted into the Army;
- When based on religious beliefs, the defense of exclusion was extended to sexual orientation (“as a Christian, I don’t want my son to have classes with a gay person”);
- Advocate for exclusion based on sex or gender from certain spaces, such as restrooms and sports leagues;
- Pronouncing insults that relate people to mental illnesses, anomalies, sins and immorality based on gender or sexual orientation.
For experts, the increased range of offenses permitted on Meta networks could increase the spread of attacks on identity groups on the internet.
“We have some areas in which we were advancing very slowly, but with this ‘carte blanche’ for hate speech, the tendency is for us to regress,” laments lawyer Natane Santos, deputy coordinator of the Advanced Studies Group on Criminal Law and New Technologies at IBCCRIM (Brazilian Institute of Criminal Sciences).
The more permissive rules come at a time when Meta's platforms were already recording increasing complaints in Brazil, according to the most recent data from the NGO Safernet.
In the case of Instagram, the total number of pages reported for crimes such as religious intolerance, racism, xenophobia, neo-Nazism and violence against women more than doubled between 2022 and 2023 (see below).
The path of investigation
Aos Fatos has identified through social media employees working for Meta, who provided us with excerpts from the company's internal documents. The guidelines were compared with the policy published by the platform and tested through denouncements.
The report also interviewed experts, consulted Safernet's survey of complaints and contacted Meta to give the company the chance to explain its standing.