Controversial social network Twitter is testing new functionality to help combat "misleading media" from AI-generated images to manipulated videos.
In a series of tweets through its @CommunityNotes account on the platform, Twitter outlined how Notes on Media will help tackle fake content.
"From AI-generated images to manipulated videos, it's common to come across misleading media. Today we're piloting a feature that puts a superpower into contributors' hands: Notes on Media. Notes attached to an image will automatically appear on recent and future matching images," the company said in a tweet.
If you're a contributor with a Writing Impact of 10 or above, you’ll see a new option on some Tweets to mark your notes as “About the image”. This option can be selected when you believe the media is potentially misleading in itself, regardless of which Tweet it is featured in. pic.twitter.com/zMtXBspN5p
— Community Notes (@CommunityNotes) May 30, 2023
Twitter is also testing out auto-generated labels, which will note if a photo or video has been altered or if it is a computer-generated image. This will help users quickly distinguish between realistic and AI-generated images, potentially helping to reduce the spread of misinformation.
While the feature is still in the early stages of development, the company is expected to launch it in hopes that it will help Twitter users make more informed decisions when sharing or viewing potentially misleading media.
Two new activities as part of Community Notes are writing and rating notes which create a rating called a Writing Impact, which indicates how often a contributor's notes have earned a status of “helpful” when rated by others. Writing Impact can increase when a message earns the status of Helpful and decreases when a note reaches the level of “not helpful”.
"If you're a contributor with a Writing Impact of 10 or above, you'll see a new option on some Tweets to mark your notes as ‘About the image’. This option can be selected when you believe the media is potentially misleading in itself, regardless of which Tweet it is featured in," Twitter said.
Twitter also said that raters and readers would see notes that authors marked as "about the image" differently, so it's clear to everyone that they should be interpreted as about the media, not the specific Tweet. Ratings can help identify cases where a note may not apply to a particular Tweet.
The feature currently supports Tweets with a single image. Twitter said it was working on expanding the features to Videos and tweets with multiple images/videos.
Recently, Twitter shared data about its health and safety efforts between January 1 and June 30, 2022. Over the reporting period, Twitter removed 6 586 109 pieces of content that violated its rules, which is an increase of 29%, during the same period, in the year prior.
IOL Tech