Yuan Announcement In January, it will end reviews of some content, relax its rules, and place more emphasis on supporting “free expression”. The company disclosed Thursday in its quarterly Community Standards Law Enforcement Report that the shift has resulted in fewer posts removed from Facebook and Instagram. Mehta said its new policy helps cut the U.S. wrong content removal in half, while not exposing users to offensive content compared to before the change.
The new report mentioned in an update to the January blog post by Meta’s head of global affairs, Joel Kaplan, which shows that metadata has dropped nearly a third of its content globally on Facebook and Instagram due to violating the January-March rules this year, less than the previous quarter’s rules, rather than 16 billion projects, rather than $1.6 billion. The tech giant’s quarterly demolition has previously risen or remained flat over the past few quarters.
In Instagram and Facebook, Meta reported that posts violating its spam rules were down by about 50%, posts that endanger children were nearly 36%, and hate behavior was down by nearly 29%. In one major rule category, deletion is only added in one major rule category (Suer and Self-harm content).
The amount of content elements fluctuates regularly from quarter to quarter, and many factors may lead to a decline in the decline. But the company itself acknowledged that “reducing changes in law enforcement errors” was one of the reasons for the massive decline.
“In a range of policy areas, we’re seeing a decrease in content volume and before user reports, we’re taking a decrease in percentage of content that we’re taking action,” the company wrote. “This is partly because the changes we’ve made to ensure we make fewer mistakes. We’re also seeing a corresponding reduction in content that attracts and ultimately recovers.”
Meta relaxed some of its content rules early in the year when CEO Mark Zuckerberg called it “distressed from mainstream discourse.” The changes have enabled Instagram and Facebook users to adopt a language that is hated by human rights activists who believe are hateful to immigrants or individuals identified as transgender. For example, META now allows for “accusations of mental illness or abnormality based on gender or sexual orientation.”
As part of the announced overhaul, just as Donald Trump was set to be the second term of the U.S. president, Meta stopped relying on automation tools to identify and delete posts suspected of violating their rules because it said their error rates were high, prompting users to be frustrated.
In the first quarter of this year, Meta’s automation system accounted for 97.4% of the company’s hate speech policy removed from Instagram, down just 1 percentage point from the end of last year. (Reporting to Meta’s users triggered the remaining percentage.) But automatic removal of bullying and harassment on Facebook dropped by nearly 12 percentage points. Meta’s system is more proactive in some categories, such as nudes, compared to the previous quarter.
