Following a troubling proliferation of AI-generated and manipulated media, Twitter announced on Tuesday its plans to expand its Community Notes system to flag altered and fake images. First launched late last year shortly after Elon Musk’s $44 billion acquisition of Twitter, Community Notes built upon the company’s previous Birdwatch program aimed at leveraging unpaid, crowdsourced fact checking of tweets to rein in misinformation and hoaxes.
[Related: Why an AI image of Pope Francis in a fly jacket stirred up the internet.]
The expansion is currently in an “experimental” testing phase, and only pertains to posts containing a single image. Twitter states it plans to extend the feature to handle tweets featuring additional media uploads such as GIFs, videos, and multiple images in the near future. As of right now, however, only those signed up as a Community Notes contributor with a user rated Writing Impact score of 10 can see the option to flag a post for its accompanying media instead of just its text. According to Twitter’s Community Notes page, “Tagging notes as ‘about the image’ makes them visible on all Tweets that our system identifies as containing the same image,” meaning that other users’ tweets containing the same image alongside different text will hypothetically contain the same flag.
Twitter’s Community Notes team warned that the new feature’s accuracy could still produce both false positives and negatives for other tweets. “It’s currently intended to err on the side of precision when matching images,” they explained, “which means it likely won’t match every image that looks like a match to you.” Twitter added that its team will continue to “tune this to expand coverage” while also cutting down on “erroneous matches.”
The new feature arrives just days after a fake image depicting an explosion at the Pentagon began circulating on Twitter, first via an account claiming association with Bloomberg News. The now-suspended account included a “Blue Checkmark” that for years reflected an account’s verified authenticity. Following Musk’s company takeover, a verification can now be obtained via subscribing to the premium Twitter Blue user tier.
[Related: Twitter’s ‘Blue Check’ drama is a verified mess.]
Twitter has relied extensively on crowdsourced moderation via the Community Notes system after axing the majority of its staff dedicated to trust and safety issues. On Wednesday, The Wall Street Journal reported the social media platform is now worth approximately one-third of the $44 billion Musk paid for it.