AI fabricated images of the Pentagon explosion causing stock market volatility, Twitter opened crowdsourcing verification

On May 30th local time, Twitter announced that it would expand its community note taking system to mark tampered and forged images. A week ago, false images of the Pentagon explosion were viral spread on the platform, and Twitter launched this new feature aimed at using free crowdsourcing to verify facts to control erroneous information and pranks.

This feature will allow users to add notes next to photos and videos on Twitter, and add context and information related to misleading content.

The community note taking feature will allow users to add notes next to photos and videos on Twitter.

The tagging function is currently in the testing stage and is only applicable to tweets containing a single image. Twitter stated that it plans to expand this feature in the near future to handle other multimedia content, such as animated images, videos, and tweets with multiple images.

Only use after reaching a certain score

A week before the announcement, an AI generated image of the Pentagon explosion was widely circulated on Twitter. This image was first sent by an account claiming to be related to Bloomberg, which has a "blue V" logo. The "Blue V" has always been a proof of account authenticity, but since Tesla CEO Elon Musk acquired Twitter, simply subscribing to the advanced TwitterBlue service can obtain this logo.

Twitter stated in its announcement that false images generated by AI have spread like viruses on the platform, some of which may cause confusion or fear among people. In some cases, these images may not pose any harm, but in the case of the Pentagon, false images had a real impact on the stock market - causing a market value change of $50 billion.

The image synthesized by AI shows an explosion near the Pentagon, triggering stock market turmoil.

Previously, the inspection of misleading and false information in community notes was limited to text. Now, contributors are allowed to add Notes on Media, which means annotating relevant information on images or videos.

Community Notes aims to create a more informative world by empowering Twitter users to collaborate and add context to potentially misleading tweets. Twitter wrote in its announcement that users' likes and reposts for tweets with community notes are 15% to 35% higher than those without notes.

Once users become contributors, they can start adding notes. To become a contributor to community notes, users must first demonstrate their ability to do so. To determine this, Twitter assigned a writing influence score to each potential contributor. The score starts from 0 and must reach 5 points to become a contributor to community notes. A score of 10 points is required to add community notes to the image. Users can improve their writing influence score by rating existing notes and other methods.

The quality of their work may result in them losing their identity as contributors, "Twitter said. In other words, if someone enters the system by following rules and then uses their identity as a contributor to amplify or spread erroneous information, they will be kicked out of the contributor team and must prove themselves again before rejoining.

Previously, comments added by community note users would be displayed to others as tweets were forwarded and cited, but if the image in the tweet appeared in other posts, the note no longer existed. Twitter stated that community notes will now appear with pictures. This means that even if a flagged suspicious image is posted elsewhere on Twitter by other users, there will still be notes next to it.

Even if a marked suspicious image is posted elsewhere by other users, there will still be notes next to it.

In many cases, these notes can provide valuable context, not only applicable to a tweet, but also to any tweet containing the same media, "wrote Twitter's official announcement.

Being questioned

However, Twitter lacks confidence in whether it can match the marked images with other similar images on the platform. The Twitter community note taking team reminds us that the new feature may encounter false positives or omissions when matching images. "We will work hard to adjust this feature to cover more images while avoiding incorrect matches

At the end of 2022, Musk launched the community note taking feature for the first time shortly after acquiring Twitter. This feature is based on the company's previous "Birdwatch" program, which aims to verify the authenticity of tweets through the use of free crowdsourcing to control the spread of erroneous information and malicious dramas.

Since Musk acquired Twitter, the content review of the platform has been a key issue. He cut most of the positions in the content review team during large-scale layoffs, but content review is the key to maintaining the normal operation of social media websites, maintaining the attractiveness and profitability of the platform.

In addition, even with complete content reviewers, Twitter finds it difficult to address the issue of false information. Trust and security experts say that distinguishing between images and videos generated by AI is what professionals are best at. Twitter wants to expand its note taking feature to involve user groups in content review, but unfortunately, these users are not known for their disciplinary or detail handling abilities.

Arjun Narayan, the director of trust and security at SmartNews in Japan, also expressed concern in an interview about this approach, saying, "This weakening of professional teams may cause trouble for the company