Twitter's Community Notes Feature Expands To Misleading Images

Amidst the widespread diffusion of misinformation through AI-generated images, Twitter has launched an experimental change that blends community participation with technology to tackle the issue. The company is testing a feature called 'Community Notes' that aims to fact-check media content, an essential pursuit in an era where manipulated images or deep fakes are becoming commonplace.

Community Notes essentially allows approved contributors with a high 'Writing Impact' score to append notes to images circulating alongside tweets, adding layers of context. For instance, these notes could indicate whether the image is AI-created or manipulated in any manner.

The ultimate goal is to implement these notes, adding context to and hopefully thereby limiting the virality of misleading content. Twitter's approach to battling visual misinformation essentially extends moderation responsibilities to the community rather than handling it internally. In its experimental phase, the Community Notes feature is limited to tweets with a single image, Twitter notes it also intends to broaden the scope to eventually include "Tweets with multiple images, GIFs, and videos."

The Community Notes process works when users link notes to images instead of specific tweets, thus allowing these notes to appear on all tweets containing the same media/imagery. Consequently, when someone rates a note, the rating is associated with the tweet that displayed the note, helping other users discern instances where a note might not align with the original content or context of a specific tweet. 

When someone marks a note as helpful, Twitter tallies the number of views that image has across all Tweets, but it only contributes to the 'Writing and Rating Impact' score of the author and raters once. Ideally, this will normalize over time as Twitter identifies sketchy accounts that post misleading images and identifies a larger pool of trusted users who are adept at spotting fake images.

Twitter is treading cautiously with Community Notes, preferring to match images with precise similarities instead of roping in images that look similar but have visually distinctive variances. Twitter says it's actively fine-tuning this process, and will also monitor feedback about the system and its efficacy overall to inform future changes.