Racist comments targeting politicians have tripled since Meta changed its rules

The researchers used an AI method trained to identify comments in the dataset that may violate Meta’s current policies in three areas: violence and incitement, hateful behavior, or oppression and harassment.

Comments that violated Meta policies related to violent threats quadrupled, from 1,800 in the six months before the change to 7,600 in the six months after. Hate comments have also quadrupled, from 6,900 to 30,000. Comments that violated Meta’s harassment and harassment policies doubled, from 15,700 to 39,900.

“We regularly report people targeting content violations on our site, and the number of hate speech has not increased in 2025,” a Meta spokesperson told WIRED, adding that the company could not directly address the report’s claims without seeing a full investigation. WIRED provided a list of the offensive words cited in the report, but Meta did not comment on it. Hours before the report was released, many of the samples were removed from Facebook.

“When companies reduce oversight in areas such as violence, hate, and harassment, it should not be surprising to see that harm increases,” Senator John Curtis, Republican from Utah and a member of the Committee on Commerce, Science, and Transportation, said in a speech to the CCDH.

Data collected by CCDH researchers are described in Meta’s your visual reports from 2025, which shows how the company reduced its compliance by half in the months following its policy change. The authors of the report wrote: “The increase in violence and the collapse of law enforcement are closely related.