When it comes to social media, we all know that in addition to the many benefits it provides in making connections easier, there can also be certain risks and dangers that come with interacting online, especially for young people. Thankfully, there are apps out there that are being proactive and trying to limit these harms that can come to young folks by implementing proper moderation technologies.

Yubo, the live social discovery app that’s most used by Gen Z, has announced that its live-stream audio moderation technology is expanding to its largest markets, namely the United States, United Kingdom, Australia, and Canada.

This comes after several months of testing the feature in some areas of the United States, as the company learned how to tackle the problems that come with analyzing audio in real time.

The Importance of Content Moderation

Moderating content online is a difficult job to do. We’ve seen this with Facebook and Twitter and Instagram and so on. Between insults, hateful comments, or even sometimes violence or self-harm, moderation is hugely important to keep these platforms safe and their users protected. There’s something about being able to hide behind the supposed anonymity offered by the internet that can bring out the worst in some people.

Well, Yubo continues to show that it is doing everything it can to help prevent this, especially since its user base is much younger. According to Yubo, 99% of their user base is Gen Z ages 13-25. While online harassment is not nice for anyone to endure, it can be particularly harmful to teens.

Image and video moderation technology has been around for a while and has been implemented with various stages of success. Audio moderation, however, has remained a major challenge, especially with live streams. Whether it’s a live stream taken while walking down the street or one where someone is playing a video game, it doesn’t really matter.

How Yubo’s Audio Moderation Works

Starting in May 2022, Yubo began a trial phase for its audio moderation technology in various areas across the United States. Now the next phase includes most English-speaking regions where Yubo has a large user base in order to allow the company to gather better insight into their new safety tool.

Although still new, this is a technology that could end up widely used to detect violence or self-harm in the future.

Yubo isn’t doing this all on its own, but rather in partnership with Hive. Hive, if you haven’t heard of them, is a company that creates and provides content moderation APIs to process images and videos in near real-time.

The technology that Yubo implemented works by recording and transcribing 10-second snippets of audio in livestreams of 10 or more people. The text is immediately analyzed by artificial intelligence to detect any words or phrases that violate the app’s Community Guidelines.

Wait, it doesn’t stop here. While artificial intelligence can be great at detecting such things, it’s still far from the point where it can differentiate between actual problematic uses and false flags.

False flags can be, for instance, songs playing in the background, or playful language that is not malintentioned.

So, this is where Yubo’s Safety Specialists come into play, as they are brought in to investigate the flagged incidents. Humans are the ones who get to determine whether a situation requires escalating or not if there are actions that should be taken against the people using bad language, or even if law enforcement needs to be brought in.

If the live stream is perfectly ok, no transcripts are reviewed or kept past a 24-hour limit, so it all remains private; or as private as a live stream with over 10 people can be.

However, if the algorithms detect violent threats, hate speech, or self-harm, for instance, the transcripts are stored for up to one year. This is, of course, a safety measure for the company in case law enforcement requires the information or the company itself needs the details for further internal investigations.

Yubo’s Chief Operating Officer, Marc-Antoine Durand, sees the expansion of audio moderation technology as a key element in the evolution of online safety across the entire industry.

“There is still a lot of progress to be made in the area of voice detection, but we are proud to be forging a path for our peers by being the first to launch audio moderation with Hive and helping make this tool more reliable and effective through this trial,” Durand said.

Obviously, social platforms don’t want rampant bullying, hate speech, and violence-inciting comments since it would push other users away. Implementing such moderation technology is a great tool to keep users safe, and also to help guarantee Yubo’s growth and success in the long run.