Study: Facebook fails to label 42 percent of debunked political misinformation
Meta’s automated content moderation and labeling systems failed to identify and tag 42% of debunked political misinformation during critical election periods, allowing demonstrably false claims to spread without context, thereby eroding voter trust and undermining democratic integrity for the general public. Researchers at online advocacy group Avaaz discovered that Facebook failed to label 42 percent of debunked political misinformation, thereby helping the spread of false narratives during the 2020 election period. The researchers calculated that Facebook could have prevented approximately 10.1 billion views on pages that disseminated misinformation had it implemented algorithm changes earlier. The top 100 pages identified as "repeat misinformers" shared an average of eight confirmed pieces of misinformation each, with many refusing to correct their posts after being flagged by fact-checkers. Facebook had only intensified its efforts to combat misinformation in the weeks leading up to the election. Prior to this, the platform allowed substantial engagement with misleading content, which included posts that were later found to be false but were not adequately addressed by Facebook's AI-powered content management system. Facebook disputed Avaaz's methodology, claiming it misrepresented the company's actions against misinformation. A spokesperson argued that the report failed to consider the context of individual pages and their content comprehensively. Nonetheless, tyhe report was seen to highlight broader concerns about Facebook's role in shaping public discourse and its responsibility in mitigating the spread of false information that can influence democratic processes. System 🤖 Facebook content moderation system Developer: Facebook Country: Global Sector: Politics Purpose: Identify, label and reduce the spread of misinformation Technology: Content modera t ion system; Machine learning Issue: Accountability; Accuracy/re liability; Human/civil rights; Transparency
- Date it happened
- 2020-10-01
- Organisation involved
- Product, system or model
- Facebook content moderation system
This incident was imported from AIAAIC and is used under CC BY-SA 4.0. Our additions to it — the structured fields, the translation, the checks against other reports — are published under the same licence.
This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.