You’ve built an engaging social platform that fosters connections between people. Unfortunately, some people would rather cause disruption and harm to others on your platform than make meaningful connections and contributions. Our context-sensing technology helps you identify and respond to bad behaviors as they unfold, across multiple languages and content types.
64% of teenagers in the United States encountered hate speech on social media in 2018. The same year the FBI reported a 16-year high in hate speech reports. There is no sign of this trend slowing down. This model identifies explicit or suggestive attacks against a person or group based on protected identity classes.
Cyberbullying can destroy lives and the fabric of an online community. This model detects the unwanted aggressive behavior.
Threats can be introduced to your community, and your revenue stream, in the blink of an eye. Your own recommendation algorithms can popularize problematic content. Spectrum operates in real time to smoothly and quickly stop inappropriate behavior.
A model is only as good as the data it is built on. It only gets better with consistent feedback. Our extensive library of trained models and classifiers is constantly updated with a wide range of trusted data sets.
Throughout the year we examine interesting topics related to Trust & Safety on social networks. These explorations help build a deeper base of knowledge to better serve communities.
Every year, on February 11th, the world celebrates #SaferInternetDay. It is a chance for us to evaluate what we’ve done to make dating apps safer, and what we're looking to do in the future.
It is a sad reality that dating apps are used to facilitate trafficking. But, there is hope. We explore what dating apps can do to prevent traffickers from using their features to find, groom and exploit victims.
Caste Discrimination, the poor treatment of certain Indian caste members, has made its way onto the Internet. We explore its history and outline ways to safeguard against it.