YouTube is enhancing its approach to curb misinformation narratives for online platforms as they continue to pose a threat in the digital era. The video streaming platform will now focus on three key areas: catching new misinformation before it becomes viral, combating the spread of misinformation between multiple platforms, and ramping up its efforts to combat hyperlocal misinformation.
For fast-moving misinformation narratives, the company is continuously updating its system with updated data and looking to leverage an even more targeted selection of classifiers and keywords in additional languages. Another challenge is the proliferation of borderline videos outside of YouTube. ‘We’ve overhauled our recommendation systems to lower consumption of inappropriate content that comes from our recommendations significantly below one percent. But even if we aren’t recommending a certain borderline video, it may still get views through other websites that link to or embed a YouTube video,’ Mohan added.
In order to address the cross-platform sharing issue, YouTube may disable the share button or break the link on videos that are already limited in recommendations. By restricting embedding and linking into borderline videos on other sites, Internet users can no longer embed or link into them. YouTube is grappling, however, with the issue of whether preventing shares may restrict a viewer’s rights too much.
Another approach could be to surface an interstitial genres before a viewer is allowed to watch a borderline embedded or linked video, informing them that it may contain misinformation. Mohan said that interstitials are like speed bumps – they make viewers pause, before watching or sharing content.
YouTube is also exploring further investments in partnerships with experts and non-governmental organizations around the globe to prevent hyperlocal misinformation, as well as expanding teams. As a result, the company is seeking ways to update models more frequently in order to catch hyperlocal misinformation, with the ability to track local languages.