Picture this: it is the night of a major American election, and you are scrolling through X, formerly Twitter, looking for real-time news. Among the flood of posts, you see a claim that election workers are shredding ballots in a swing state, or that certain cities have quietly made voter ID illegal. The posts are angry, confident, and shared by thousands of people. But there is no context beneath them, no note saying “this is false,” no link to official sources. For most users, these posts simply look like truth. That scenario is not hypothetical. A new report from NewsGuard, a US watchdog that monitors the credibility of news and information, has found that X’s crowd-sourced fact-checking system, Community Notes, failed to flag the vast majority of false claims related to the US midterm elections. Out of dozens of misleading posts that were identified by researchers, only 13 carried any kind of correction. That means 93.1 percent of the false claims examined went completely unflagged, even as they racked up millions of interactions. The findings are a blunt reality check for a platform that has repeatedly positioned itself as a place where misinformation is handled by the people, for the people.
Community Notes was introduced on X in late 2022, shortly after Elon Musk took over the platform. The idea was simple and, in many ways, appealing. Instead of relying on professional fact-checkers, X would allow everyday users to write short contextual notes that could be attached to misleading posts. Other users would then rate whether those notes were helpful. If enough people from different perspectives agreed, the note would publicly appear beneath the post. Musk has often promoted this feature, calling it “awesome” and insisting that it makes the platform more democratic. The model has since been embraced by other tech companies, especially as they look for ways to reduce their own responsibility for content moderation. But the NewsGuard study suggests that Community Notes is far from the safety net it claims to be. The false claims that slipped through were not subtle. They included a viral post from a conservative commentator alleging that New York and California had made it illegal for voters to show their ID, and illegal for election officials to ask for it. That is entirely made up. No such law exists in either state. Yet the post spread without any correction attached to it, reaching users who had no reason to doubt its accuracy.
Perhaps the most striking part of the report is what it says about Musk himself. The billionaire, who has more than 241 million followers on X, has repeatedly said that Community Notes applies equally to everyone, including him. He has framed the feature as a kind of equalizer, a way to ensure that even the most powerful voices can be publicly corrected when they are wrong. But NewsGuard found that Musk personally advanced at least four false claims related to the midterm elections, and none of those posts carried a Community Note. That is a powerful irony. If the system is supposed to be so effective, why does it fail to work precisely when the platform’s most visible user spreads misleading information? It is possible that the notes are being written and submitted, but perhaps they are not being approved quickly enough, or perhaps they are being rated down by people who disagree with them. It is also possible that the sheer volume of engagement around Musk’s posts makes it harder for corrections to gain visibility. Either way, the result is the same: millions of people were exposed to falsehoods without context, and the platform’s own safeguards did not step in. X did not respond to the report, leaving those questions unanswered.
Researchers have been warning for some time that Community Notes has serious limits. Studies have shown that it can be effective in certain narrow cases, such as debunking well-established vaccine misinformation, where there is broad scientific agreement and little political controversy. But the system struggles with issues that are politically polarized, which is precisely where misinformation tends to thrive during elections. A false claim about voter fraud may be easily debunked by official records, but if the claim is being shared by a popular political figure, those who are inclined to support that figure may not rate the corrective note as helpful. In fact, they may actively rate it down. This creates a strange dynamic in which truth becomes a popularity contest. A correction only appears if enough people from across the political spectrum agree on it, but in a highly polarized environment, that kind of agreement is rare. Some researchers have also warned that Community Notes can be gamed by organized groups who coordinate to target political opponents. Instead of being a neutral crowd-sourced system, it can become another battleground for partisan warfare, where users are motivated not by accuracy but by loyalty to their side.
The implications go far beyond X. Meta, the parent company of Facebook and Instagram, has also been moving in this direction. Last year, Meta ended its third-party fact-checking program in the United States, a move that its CEO, Mark Zuckerberg, defended by claiming that professional fact-checkers had led to “too much censorship.” Professional fact-checkers strongly rejected that accusation, but the decision still stood. More recently, Meta announced that it had begun testing its own version of Community Notes in sixteen Spanish-speaking countries across Latin America, as a step toward fully replacing its traditional fact-checking model. No timeline has been given for when this change will become official, but the direction is clear. Tech companies are increasingly eager to hand the responsibility of verification to users, not because it works better, but because it is cheaper and removes the platform from the uncomfortable position of deciding what is true. Yet the NewsGuard study shows that this approach has real consequences. False claims that are not corrected can influence public opinion, undermine trust in democratic institutions, and even lead to threats of violence against election workers. The fact that these claims are allowed to circulate without restraint is not a neutral outcome. It is a decision to prioritize engagement over accuracy.
Even Meta’s own Oversight Board, a quasi-independent body funded by the company to review its moderation decisions, has cautioned that Community Notes is not an adequate replacement for professional fact-checking. The board has pointed out that crowd-sourced systems are slow, inconsistent, and vulnerable to manipulation. They are especially unreliable in breaking news situations, when misinformation spreads rapidly and users are still trying to understand what is happening. By the time a note is proposed, rated, and approved, the damage may already be done. This is not an argument for dismissing Community Notes entirely. It can be a useful supplementary tool, and there are examples of good corrections being written and featured. But the report is a reminder that there is no easy substitute for professional journalism and dedicated fact-checkers. AFP, the news agency that reported this story, currently works in 26 languages with Meta’s global fact-checking program, including across Latin America. That kind of human effort, with editors who can verify documents, interview officials, and trace the origins of a false claim, cannot be replaced by a comment box. As the midterm elections showed, the stakes are too high for us to pretend otherwise. The next time you see a viral claim on social media, ask yourself whether it has been verified. More importantly, ask yourself who benefits when we stop checking. Community Notes may be a nice idea, but it is not a substitute for the truth.