Facebook Creates New AI To Detect Hate Speech, Offensive Memes

Facebook introduced another tool in its fight against spam and hate speech this week in the form of a new AI system codenamed “Rosetta.” The system is designed to read the text found in videos and images on the platform. There are obviously tons of uses for this technology, but it will most likely serve a major role detecting hate speech and offensive content.

It’s not uncommon to find some of the darkest content on Facebook embedded in memes or images. The same goes for videos. And as the technology to fake video content becomes more widely available, Facebook will want to be at the forefront of stopping it. According to the company, Rosetta will also be used to surface content in your News Feed and in photo searches.

This rollout comes in the middle of the social media giant facing intense public scrutiny over its privacy and content moderation practices. Just last month, the company announced it will now remove disinformation intended to “spark or exacerbate” violence, including videos and images. Previously, the platform only banned content that explicitly called for violence.

This is definitely a step in the right direction for Facebook, but it’s not yet clear how effective it will be. After all, the site relies heavily on AI now, and plenty of phony and hateful content still sneaks through.