Meta’s Community Notes System Faces Serious Challenges in Global Expansion

Meta’s Oversight Board has released a comprehensive analysis raising significant concerns about the company’s planned global expansion of Community Notes, concluding the system has fundamental weaknesses that make it vulnerable to manipulation and ineffective at combating misinformation.

According to the Board’s findings, Community Notes – which Meta implemented at the beginning of Donald Trump’s second term as a replacement for professional fact-checking – suffers from extensive delays in publication, limited coverage, and susceptibility to coordinated manipulation, especially in the face of AI-generated disinformation.

“Delays in note publication, the limited number of published notes and its dependence on the broader information environment’s reliability raise serious doubts about the extent to which community notes can meaningfully address misinformation linked to harm,” the Board stated in its analysis.

The scale disparity between Community Notes and professional fact-checking is stark. During the first six months of the U.S. rollout, Meta published approximately 900 Community Notes. In contrast, professional fact-checkers in the EU enabled Meta to apply labels to roughly 35 million Facebook posts during a comparable period, according to Angie Drobnic Holan, director of the International Fact-Checking Network.

Meta’s system, built on X’s (formerly Twitter) open-source algorithm, allows users to propose contextual annotations on public posts, with other users rating them as “helpful” or “not helpful.” Only notes that achieve consensus among typically disagreeing users are published. The bar for publication is high – just six percent of proposed notes ever get published, slightly below X’s 8.3 percent publication rate.

A critical difference from Meta’s previous fact-checking program is that posts receiving Community Notes face no penalties. Content is neither downranked nor excluded from recommendations, and there are “no strikes for posting content that receives a community note,” the Board observed. The previous fact-checking system would demote false or misleading content in distribution and reject it for monetization.

The Board expressed particular concern about AI-powered manipulation, warning that “artificial intelligence facilitates the scaled creation and operation of accounts and networks” that could game the system. Research cited by the Board shows that even a small minority (5-20%) of coordinated bad actors can strategically suppress helpful notes, while published notes remain vulnerable to coordinated negative rating campaigns for up to two weeks.

While Meta told the Board it prohibits AI note writers and has not detected any coordinated manipulation to date, the Board questioned whether these safeguards would be adequate as AI tools become more sophisticated.

Crisis situations dramatically highlight the system’s limitations. Examining the 2024 Southport riots in the UK, the Board found that five accounts spreading false information generated over 430 million views. Of the 1,060 posts from these accounts during the height of the riots, only one received a community note.

Based on these findings, the Board recommends against introducing Community Notes in countries experiencing crises or protracted conflicts, citing Meta’s previous failures in content moderation that contributed to offline violence in Myanmar and Ethiopia.

The system’s algorithm also presents structural weaknesses by modeling societal polarization along a single axis. In countries with complex, overlapping divisions spanning ethnicity, religion, language, and caste, this approach could systematically disadvantage minorities. The Board cited evidence from a consortium of South Asian NGOs demonstrating this exact dynamic in X’s Community Notes implementation in India.

Language disparities pose additional challenges. The system currently operates in just six languages, with research showing non-English notes on X are rated and published far less frequently than English ones.

For its planned international expansion, the Board recommends a carefully staged rollout with strict exclusion criteria. Countries with repressive human rights records, active conflicts, histories of coordinated disinformation, or upcoming major elections should be excluded until Meta can demonstrate robust safeguards.

Importantly, the Board emphasized that community notes and professional fact-checking “should not be seen as mutually exclusive tools.” Research shows community notes frequently cite fact-checking sources, suggesting that reducing support for professional fact-checkers could undermine the quality of the Community Notes system itself.

While Meta is not legally required to comply with the Board’s recommendations, the analysis presents a clear warning about the limitations of crowd-sourced fact-checking in an era of increasingly sophisticated AI-generated misinformation.