Can Community Notes match the speed of misinformation?
As Meta scraps fact-checkers for Community Notes, AI-boosted Supernotes hold promise
What happens when platforms stop fact-checking? With Meta’s decision to replace fact-checkers with Community Notes, the question isn’t just whether the new system can work—but whether it can keep up with the speed of misinformation. That’s where recent experimentation with AI-enhanced "Supernotes" could be an intriguing next step.
What are Community Notes?
Community Notes, originally implemented by X, is a crowdsourced system where users flag misleading posts and provide context. These notes are rated by others on their helpfulness, with bridging algorithms surfacing the notes that receive approval from diverse groups of users. This "bridge-based ranking" ensures that only notes approved by users across different perspectives are shown, making it more credible.
The promise of Community Notes lies in its transparency and its ability to crowdsource moderation from across ideological divides. By emphasizing consensus, the system avoids the mistrust or perception of bias with platform-driven fact-checking or content removal. Last year YouTube adopted this approach, but as a complement to other products such as information panels, or their recent disclosure requirement when content is altered or synthetic.
What the research tells us
Community Notes is labelled as ‘convincing’ in reducing the spread of misinformation, according to research curated on Community Notes by the ProSocial Design Network library, a curated collection of evidence-based design patterns and tools for creating healthy online spaces.
Wojcik et al. (2022): Twitter’s internal field experiment found that adding Community Notes to misleading tweets reduced retweets by 25–34%.
Chaui et al. (2023): By analyzing Community Notes’ staggered rollout, researchers found that tweets with visible notes saw a decrease in retweets after the U.S. rollout. The researchers also found ‘Community Notes might be too slow to effectively reduce engagement with misinformation in the early (and most viral) stage of diffusion.” In addition, no similar effect was observed during the global rollout.
Renault et al. (2024): This study showed that tweets with visible notes saw a 50% decrease in retweets, and showed an increase by 80% the probability that a tweet is deleted by its creator when a Community Note was added.
Chuai et al. (2024): This study reported a 60% reduction in retweets for tweets with visible notes, but also suggested that community notes might be too slow to intervene in the early (and most viral) stage of the diffusion.
The critical limitation is timing. Notes typically become visible 15 hours after a tweet is posted, according to Chuai et al, by which time 80% of retweets have already occurred. As a result, the overall reduction in misinformation reshares across all tweets is modest—estimated at around 11% .
AI and the Supernotes
To address the timing and scalability issues of Community Notes, researchers recently tested an AI-enhanced system called Supernotes. This system leverages Large Language Models (LLMs) to rapidly synthesize the most helpful elements of multiple user-generated notes into a single, high-quality “Supernote.” The experiment showed several win themes:
Speed: Published within 2 hours on average, compared to 15 hours for Community Notes.
Quality: Participants rated the Supernotes higher than human-generated notes; when asked why, researchers found it was because, ‘‘the Supernotes effectively synthesize information from multiple notes, providing more holistic context, including more sources, and presenting the information in clear, unbiased, and non-argumentative language.”
Impact: Tweets with Supernotes saw a reduction in retweets similar to—or slightly better than—traditional Community Notes, but the faster publication time allowed for a more significant overall impact.
Scalability: Automated systems cover more content than traditional notes.
With both the Community Notes approach or an eventual Supernotes-type deployment, there are risks to consider, including:
Trust and Transparency: Platforms must ensure that the AI-generated notes are not perceived as biased or opaque. Open datasets and explainable AI methods could help build trust.
Volunteer Diversity: Community Notes, and by extension Supernotes, rely on diverse user input to function effectively. If the volunteer base is ideologically skewed, or gamed, the system's ability to produce unbiased notes is compromised.
Cost and Feasibility: Scaling Supernotes to cover all relevant content may require significant computational resources and human oversight, raising questions about long-term sustainability.
Meta’s pivot toward Community Notes marks a turning point. Misinformation spreads rapidly and can fuel polarization and trigger real world violence. As Meta embarks on its transition to Community Notes, it’s worth asking: can these tools meet the challenge, or will platforms need to adopt even more innovative solutions to tackle the persistent problem of misinformation?
Certainly, there is a lot more to unpack behind the entirety of Meta CEO Mark Zuckerberg’s announcement. What are your thoughts about this move to Community Notes? What might be the effect of dropping of fact checking on a global level ? We’d love to hear your thoughts.
Lena Slachmuijlder is Executive Director of Digital Peacebuilding at Search for Common Ground, and Co-Chairs the Council on Tech and Social Cohesion
Great post. Nice to learn AI can improve speed!
Another big question is coverage. One weakness of community note on twitter is that it’s only flagging a tweet when both republicans and democrats agree that it should be flagged (I’m simplifying a bit here).
This is fine to detect scams and very obvious forms of misinformations that have no political implications, but when a topic is politically-loaded, the community note system is much less likely to flag it. So I see a risk that divisive politicians will basically get a free pass to say whatever they want, even when factually untrue, just because one side of the isle likes them enough to close their eyes or forgive the lie. So at the end of the day, the switch to community notes might catch more miss-information overall, but also allow more political propaganda.