January 2025 reporting on leaked Meta moderator-training examples described some degrading statements about groups as permissible under revised content rules—but it does not establish that users can post every example in every context or guarantee how moderators will enforce a rule. One example involving a 17-year-old was later identified by Meta as a mistake that was meant to show content that was not allowed.
What the leaked-document reporting says
Futurism reported on January 10, 2025, that examples in training material for Meta’s revised content rules treated certain demeaning statements about groups—including statements concerning race, religion, immigration and sexual orientation—as allowable. Futurism attributed its account of the examples to The Intercept’s reporting. The underlying training document and The Intercept article are not directly available in the material supporting this account, so the examples should be understood as reported descriptions, not independently verified quotations from Meta’s document. Read Futurism’s January 10, 2025 report.
The headline’s “racism dial” is a metaphor, not the name of a Meta control, formal metric or published policy tool. Futurism connected the phrase to a Dril tweet.
One reported example was corrected
Futurism also reported that Meta told The Intercept an example involving a 17-year-old had been included by mistake and was intended to illustrate content that was not allowed. That correction matters: the example should not be read as evidence that Meta’s guidance permitted that content.
Free tools Windows power users keep installed
One-click scans. No signup required.
What the examples do—and do not—establish
Moderator-training examples can illuminate how a policy boundary is interpreted, but they are not an exhaustive public rulebook. The reporting does not establish that every user may post each described statement in all circumstances, that a post will remain online, or that enforcement outcomes are guaranteed. Context and the policy version in force matter, and the available account does not provide a region-by-region or language-by-language breakdown.
The report is a snapshot from January 2025, not confirmation of Meta’s current Community Standards. It supports a limited conclusion: the reported training examples raised questions about where the company drew boundaries for demeaning speech, while one example was explicitly described by Meta as an error.
Rank #2
- Keep track of everything from attendance to test scores
- Spiral bound
- Measures 8-1/2" x 11"
Why Meta’s hate-speech statistics do not settle the question
Meta’s published statistics measure different things from the examples in the 2025 report. In October 2021, Meta vice president of integrity Guy Rosen said hate speech appeared in about 0.05% of content people viewed—roughly five views per 10,000—and that this prevalence measure had fallen by almost 50% over the preceding three quarters. Meta defined prevalence as the amount of hate speech people actually see, including the effects of reducing distribution, rather than relying only on removal counts. These were company-reported figures for 2021, not an independent audit or a measure of how often the later examples were allowed or enforced. Meta’s explanation of its 2021 hate-speech measurement.
Other 2021 figures use different denominators and have different attributions. The Washington Post reported an internal Facebook estimate that the company removed less than 5% of all hate speech; Meta spokesperson Dani Lever disputed that characterization, saying the selected documents lacked context. The Post also explained that Mark Zuckerberg’s 94% figure referred to the share of hate speech the company removed that it had detected before a human report—not 94% of all hate speech on Facebook. Neither figure directly confirms or refutes the specific examples reported in 2025. The Washington Post’s Facebook Papers explainer.
Internal moderation guidance has drawn scrutiny before
In 2017, The Guardian reported on leaked Facebook moderation guides and the questions they raised about transparency and how moderators interpreted rules. That reporting concerned documents and policies from an earlier period; it is useful historical context, not proof that the 2017 guidance remained in force in 2025. The Guardian quoted Facebook policy chief Simon Milner discussing staffing and moderation operations, and reproduced guidance that framed online dialogue as reflecting offline exchanges. The Guardian’s 2017 report on Facebook moderation guides.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
