What is the best way to fight misinformation online? Up until recently, most companies decided to label content with third-party fact-checkers, but that approach was limited and top-down. Now, with X recently expanding its Community Notes program—even integrating AI to scale it— the platform is betting on a more decentralized, user-driven model to handle misinformation, which is a major shift in online fact-checking. Consumers will be the ultimate beneficiaries.
In the not-so-distant past, companies such as Meta and X chose third-party fact-checking as their preferred method to crack down on misinformation. Meta rolled out its third-party fact-checking program back in 2016, and X, then Twitter, followed suit with their own in-house fact-checking in 2020.
As Harvard and MIT studies show, this approach has been effective in discouraging the spread of misinformation. However, many consumers quietly felt that their content was being purposefully censored on social media—a confirmed intuition verified by Meta CEO Mark Zuckerberg’s admission that Facebook would censor social media content at the behest of the federal government. Most importantly, though, is that there has never been an improvement in how this third-party fact-checking is done. Today, many of the same criticisms remain unaddressed. An R Street Institute literature review on fact-checking finds that political bias was present when reviewing partisan content online, especially conservative content, which explains the skepticism from conservatives towards third-party fact-checking.
The truth is that fact-checking is difficult. Verifying narrow claims is much easier than fact-checking more complex issues that require context and nuance. Those are the claims that often lead to mislabeling. Given the sheer number of posts needing fact-checking, the volume of information can quickly exceed the capacity of fact-checkers, which makes this method difficult.
Fortunately, there’s an alternative: letting users provide direct feedback on posts. This approach was first introduced by Twitter in 2021 through Birdwatch, a program that allowed users to add context to potentially misleading content. Birdwatch eventually evolved into Community Notes, with the key distinction from traditional fact-checking that it emphasizes a diversity of perspectives contributing to an ongoing conversation. The entire process is much more collaborative.
A growing body of evidence suggests this model works. A study from Cornell University found that Community Notes cut the likelihood of retweeting a flagged post by half and made authors 80 percent more likely to delete the post. Another study showed that users often found Community Notes more trustworthy than fact-checks, increasing consumer satisfaction online, largely because community notes offer context rather than simply labeling something as false. In other words, community notes are much more equipped to grapple with nuance—and it’s not just social media that leverages this approach. Wikipedia, for example, also relies on the idea of crowdsourcing. Community notes bring a variant of this style to social media.
Still, this isn’t a silver bullet. Critics have raised valid concerns about the system’s limitations. For instance, a Bloomberg analysis found that Community Notes often arrive too late to prevent misinformation from spreading—an issue echoed by the Cornell study. Many notes never get published at all—not because they are inaccurate—but because they fail to reach the required level of cross-ideological consensus. This highlights the slow pace at which these notes are published.
But comparing crowdsourcing with fact-checking inherently means evaluating how these different approaches address flaws in their fact-checking process, and this is where the Community Notes approach is clearly superior. We are seeing an evolution from Community notes to meet changing market needs and conditions better than the third-party approach. In a recent announcement, X revealed plans to use AI to help draft Community Notes—an effort to expand their reach and accelerate review times, addressing volume issues that hindered earlier third-party fact-checking models. While human oversight is still required, the company is now training AI systems using human feedback, laying the foundation for a more scalable, hybrid system. This AI integration is fundamentally different from X’s chatbot, Grok. Rather than generating conversational responses, the goal here is to enhance collaborative fact-checking, not simulate dialogue with users directly.
But perhaps most exciting is how quickly other platforms are adopting similar approaches. Just this year, Meta (across Facebook, Instagram, and Threads), TikTok, and YouTube have introduced features resembling Community Notes. If X succeeds in implementing its latest AI tool in this process, why wouldn’t other companies follow this improved version? Taking it even a step further, nothing prevents other social media platforms from releasing their own improvements that other companies will try to match and improve, so the process of competition and innovation continues. This kind of competition simply didn’t exist to the same degree in third-party fact-checking.
In the end, the biggest winners are the consumers who can now use better products with more nimble fact-checking arrangements, while also gaining a more meaningful role in shaping online discourse. That just feels like it fits the American values of free expression more closely. Of course, Community Notes isn’t perfect and will still have issues in the future. But as social media platforms expand their use of crowdsourcing, there’s nothing stopping them from refining this model further. This is just the beginning. The free-market fight against misinformation in favor of free online expression is just beginning.
Pablo Garcia Quint is a technology policy analyst with the American Consumer Institute, a nonprofit education and research organization. For more information about the Institute, follow us on X @ConsumerPal.