Functional Tests For
Hate Speech Detection Models

Get targeted diagnostic insights into model performance


This website is your first port of call for all things HateCheck. HateCheck provides targeted insights into the performance of hate speech detection models. By revealing model strengths and weaknesses, it supports the creation of fairer and more accurate hate speech detection models.

How HateCheck Works

HateCheck is made up of different functional tests. Each test evaluates model performance on a specific type of hateful or non-hateful content. HateCheck has now been expanded to cover 11 languages, with 25+ functional tests in each language, and special tests for emoji-based hate in English.

Media Coverage

HateCheck has been featured in major news outlets around the world, from MIT Technology Review to Bloomberg News. Visit our newsroom or check out our blog to find out more.

HateCheck Support

Get in touch with the HateCheck team via the button below.
We welcome your feedback and questions!