As AI-generated images and videos become more prominent on Twitter, the company is testing a new feature that could make it easier for people to identify potentially “misleading” media. The company is piloting community feedback for media, which will apply validation checks from the audience to specific photos and videos.
This feature allows Community Notes contributors with high enough ratings to apply notes to images shared in Tweets. Like notes on tweets, labels can add additional “context” to images, such as indicating whether an image was generated using synthetic AI or otherwise manipulated.
From AI-generated images to manipulated videos, it’s common to come across misleading media. Today we’re experimenting with a feature that puts a great power in the hands of contributors: media feedback
Photo notes will automatically appear on recent and future matching photos. pic.twitter.com/89mxYU2Kir
CommunityNotes May 30, 2023
The feature could also address the viral spread of such images. According to Twitter, the goal is for notes to automatically appear on “recent and future” versions of the same photo even if it’s been shared by separate users in new tweets. However, Twitter notes that it will take some time to perfect image matching. “It is currently intended to err on the accuracy side when matching images, which means that you will likely not match every image that looks identical to you,” . “We will fine tune this to expand coverage while avoiding false matches.”
It should also be noted that Community Notes’ track record is far from perfect. While the feature can sometimes lead to rigorous fact-checks or debunking of false claims, Community Note contributors themselves have done just that. that the feature “is not impervious to errors or perpetuates common misconceptions”.
Right now, Twitter is only testing media notes for tweets with a single image, but the company says it plans to expand the feature to tweets with multiple images and videos in the future. Twitter is not the only platform grappling with how the rise and spread of artificial intelligence is generated. Google also recently has features that help users track an image’s search history, which could help searchers know whether or not an image is fake.