Last year, Meta radically overhauled the rules around what content it would allow on its platforms, claiming its own policing efforts had gone too far. Over a year later, new research from the Center for Countering Digital Hate (CCDH) shows the immediate impact of these changes: threats against politicians skyrocketed.
Meta's Policy Overhaul
“We have been over-enforcing our rules, limiting legitimate political debate and censoring too much trivial content,” Joel Kaplan, Meta’s chief global affairs officer, wrote in a blog post at the time. The company relaxed its rules, but according to CCDH, the result was a surge in abusive content.
A Meta spokesperson told WIRED: “We regularly issue public reports tracking violating content on our platforms, and the prevalence of hateful conduct did not increase throughout 2025.” The company did not comment on specific abusive comments cited in the report, many of which were deleted from Facebook hours before the report was published.
CCDH Research Findings
CCDH researchers analyzed about 8 million Facebook comments from 100 members of the House of Representatives—50 Republicans and 50 Democrats with the most Facebook followers. They scraped comments from six months before and after Meta’s policy changes, using an AI system to identify violations in three areas: violence and incitement, hateful conduct, and bullying and harassment.
Key findings include:
| Category | Before Changes | After Changes | Increase |
|---|---|---|---|
| Violent threats | 1,800 | 7,600 | quadrupled |
| Hate speech | 6,900 | 30,000 | quadrupled |
| Bullying/harassment | 15,700 | 39,900 | doubled |
Abusive and racist comments targeting both Republican and Democrat lawmakers tripled overall. Threats against President Trump more than doubled, with some comments that could be classified as felony offenses. The report cites specific examples of gendered and racist abuse directed at lawmakers like US representatives Jasmine Crockette of Texas and Byron Daniels of Florida, which were not taken down by Meta.
The data echoes Meta’s own transparency reports from 2025, which show the company cut its proactive content moderation enforcement by roughly half after the policy changes. “The surge in abuse and the collapse in enforcement track one another almost exactly,” the report’s authors write.
Industry and Political Reaction
“When companies reduce oversight in areas like violence, hate, and harassment, it should not be any surprise to see those harms increase,” Senator John Curtis, a Republican from Utah and member of the Committee on Commerce, Science, and Transportation, said in a statement to CCDH.
Nina Jankowicz, CEO of the American Sunlight Project, who briefly led the Disinformation Governance Board under President Joe Biden, noted that “threats and abuse perform well, as do the responses to the threats and abuse,” explaining the platform incentives.
The research highlights a tension between free speech principles and the operational reality of content moderation at scale. While Meta’s Kaplan framed the changes as correcting over-enforcement, the CCDH data suggests the pendulum swung too far.
Implications for Trust and Safety
For enterprise technology leaders, the case underscores the risks of algorithmic content moderation and the importance of transparent enforcement metrics. The CCDH methodology—using an AI system to detect violations in a dataset of 8 million comments—demonstrates how automated tools can audit platform behavior, though the Meta spokesperson disputed the prevalence of violations.
The findings also have broader relevance for any organization managing large-scale user-generated content: policy changes must be measured against real-world harm, not just internal metrics. The collapse in enforcement tracked by Meta’s own reports suggests that without proactive oversight, abusive content can quickly overwhelm systems.
Sources:
