Research results show that community notes do not function adequately against misinformation spread by verified accounts of X (Twitter)
Elon Musk acquired Twitter and changed its name to 'X,' limited the blue check mark indicating verified to subscribers, and allowed third parties to add supplementary information to posts. We have adopted various specification changes, including the introduction of the 'Community Note' function. Non-profit media ProPublica and Columbia University's Tow Center for Digital Journalism have published research showing that verified accounts are spreading misinformation and that ``community notes'' are not working well.
Verified Accounts on X Are Thriving As They Spread Israel-Hamas Conflict Misinformation — ProPublica
Verified Accounts, 'Community Notes' Perpetuate Misinformation On X, Study Finds 12/22/2023
https://www.mediapost.com/publications/article/392118/verified-accounts-community-notes-perpetuate-mi.html
A research team from ProPublica and Columbia University's Tow Center for Digital Journalism looked at how misinformation about the conflict between Israel and Hamas, which began in October 2023, spiked on X between October and November 2023. I investigated. The research team investigated claims found to be misleading through independent fact-checking and scrutinized posts from verified accounts posting them.
``Posting misinformation'' was narrowed down to posting photos and videos of unrelated incidents, video of games, and images generated by image generation AI, along with false content and one's own biased claims. . More than 2,000 posts were investigated, and more than 200 of them were found to have false content on X. They were published by over 1,300 verified accounts, and the total number of post impressions reached 500 million.
According to the research team, the number of verified accounts that posted misinformation increased from October to November 2023, with some posting misinformation three or more times and having more than 100,000 followers. There were apparently dozens of accounts.
Below is a graph showing the changes in the number of followers of verified accounts that post posts containing incorrect information (orange), Israeli media (blue), and Palestinian media (green). While the number of followers of Israeli media and Palestinian media has hardly changed, the number of followers of verified accounts has rapidly increased, with one person increasing five times to more than 2 million in about two months. doing.
The fastest-growing account mentioned above belongs to a 24-year-old self-proclaimed political commentator who calls himself a '
According to the research team, 80% of the 2,000 posts investigated were not corrected by community notes, and more than 80 of the 200 posts on X were found to contain false information. The error was discovered outside of the community notes.
Below is an example of a post about the conflict between Israel and Hamas with the same image generated by AI attached. Of the four posts, the three on the right have a community note with the correction, ``This is an image generated by AI,'' but the leftmost post with the oldest posting date and time does not have a community note attached. not.
Additionally, an analysis of 400 misinformation posts posted within two weeks of the initial attack on October 7, 2023 found that on average it took 7 hours for a community note to appear after a post was posted. It turns out that it took. Furthermore, even in the first week of November, it took more than 5 hours for the community note to appear in the post.
Regarding Community Notes, when we provided the survey results and the collected dataset to Keith Coleman, head of the Community Notes team at This is just a small portion of the community notes that were posted.Also, there were many cases where posts with highly visible community notes were deleted, including some that we were unable to identify.'' Furthermore, regarding a post that attached an image generated by an image generation AI to a text about the conflict, Coleman said, ``The user did not realize that the image was generated by an AI, and it was not thought that a community note was necessary.'' There's a possibility.' I answered.
'Community Notes was upgraded in response to misinformation on 'It means that we are doing well,' he said.
However, Mike Anany, an associate professor of communication journalism at the University of Southern California, said he is skeptical about leaving moderation up to the community. 'Given all the trends that have been shown, I don't think any community norms will work.'
Related Posts:
in Web Service, Posted by log1i_yk