Over two billion people use Facebook every month, which can make it difficult to monitor content posted to the social media platform.
To help with this, Facebook has created a machine learning system known as Rosetta, which is capable of scanning over one billion posts on Facebook and Instagram every day.
In a new blog post, Facebook said Rosetta can be used to “automatically identify content that violates our hate-speech policy.”
To do this, the artificial intelligence detects rectangular regions in images that normally contain text and translates the data into information that is readable by machines. Rosetta can then transcribe this data into different languages, including those like Arabic that don’t use English or non-Latin alphabets.
Facebook also announced that its automatic translation services now support 24 new languages, including Nepali, Punjabi, Serbian, Somali and Urdu.