Next week, Meta will begin one of the company’s most significant overhauls ever for how it fact-checks information on its platforms.
On March 18, Meta will start releasing its version of Community Notes for Facebook, Instagram, and Threads users in the United States. The program copies a crowdsourced fact-checking system that Twitter unveiled in 2021 and that became the sole means of correcting misleading information after Elon Musk turned the platform into X.
Meta executives say they’re focused on getting Community Notes right in the U.S. before it rolls the feature out to other countries. It’s a high-stakes region for testing a major new feature, given that the U.S. is Meta’s most lucrative market, but Meta may be hesitant to roll out Community Notes in other regions such as the European Union, where the European Commission is currently investigating X over the effectiveness of its Community Notes feature.
The move could also signal Meta CEO Mark Zuckerberg’s eagerness to appease the Trump administration, which has previously criticized Meta for censoring conservative viewpoints.
Zuckerberg first announced these changes in January as part of a broader effort to give oxygen to more perspectives on his platforms. Since 2016, Meta has relied on third-party fact checkers to verify information on its platforms, but Neil Potts, Meta’s VP of Public Policy, told reporters in a briefing on Wednesday that the systems were too biased, not scalable enough, and made too many mistakes.
For example, Potts said Meta applied false fact-checking labels to an opinion article on climate change that appeared in Fox News and the Wall Street Journal. In another case, Zuckerberg recently said on Joe Rogan’s podcast that Meta should not have dismissed concerns around COVID-19 vaccines as misinformation.
Meta hopes that Community Notes will address public perception that it’s biased, make fewer mistakes, and present a more scalable fact-checking system that ultimately addresses more misinformation. However, Meta notes this system does not replace Community Standards – the company’s rules that dictate whether posts are considered hate speech, scams, or other banned content.
The overhaul of Meta’s content moderation comes at a time when many tech companies are trying to address historical biases against conservatives. X has led the industry’s effort, with Elon Musk claiming to center his social platform around “free speech.” OpenAI recently announced it was changing how it trains AI models to embrace “intellectual freedom” and said it would work to not censor certain viewpoints.
Rachel Lambert, Meta’s Director of Product Management, said in the Wednesday briefing that Meta is basing its new fact-checking system off of X’s open-source algorithms around Community Notes.
Meta opened applications for contributors to its Community Notes network in February. Meta’s contributors will be able to suggest notes that directly fact-check claims in a post on Facebook, Instagram, or Threads. Other contributors will then rate a note as helpful or not helpful, determining in part whether the Community Note will appear to other users.
Much like X’s system, Meta’s Community Notes system evaluates which contributors normally disagree on posts. Using this information, Meta will only display a note if sides that typically oppose each other agree that a note is helpful.
Even if a majority of Meta’s contributors believe a Community Note is needed, it doesn’t mean one will be shown. Further, Meta says it won’t downrank a post or account in its algorithms even if a community note is shown on a post.
For years, crowdsourced systems like Community Notes have been seen as a promising solutions to address misinformation on social media, but they have drawbacks.
On the positive side, researchers have found that people tend to see Community Notes as more trustworthy than flags from third-party fact checkers, according to a study published in the journal Science.
In another large scale study on X’s fact-checking system, researchers with the University of Luxembourg found that posts with Community Notes attached to them reduced the spread of misleading posts by 61%, on average.
But a lot of posts don’t get notes attached to them, or it takes too long. Because X, and soon Meta, require Community Notes to reach a consensus among contributors with opposing viewpoints, it often means that fact-checks are only added after a post has reached thousands or millions of people.
The same University of Luxembourg study also found that Community Notes may be too slow to intervene in the early and most viral stage of a post’s lifespan.
A recent study from the Center for Counseling Digital Hate highlights the conundrum. Researchers took a sample of posts containing election misinformation on X and found that contributors suggested accurate, relevant information on these posts 81% of the time.
Yet of those posts that received suggestions, only 9% received consensus among contributors, meaning that a large majority of these posts didn’t appear with any fact checks.
Maxwell Zeff is a senior reporter at TechCrunch specializing in AI and emerging technologies. Previously with Gizmodo, Bloomberg, and MSNBC, Zeff has covered the rise of AI and the Silicon Valley Bank crisis. He is based in San Francisco. When not reporting, he can be found hiking, biking, and exploring the Bay Area’s food scene.