AI Solutions for Online Safety
HateCheck.ai was created by a team of researchers at Rewire, a tech startup building socially responsible AI for online safety. The project was supported by Google’s Jigsaw team.
About The Creators Of HateCheck
HateCheck has been developed by AI researchers and online safety experts working on toxic language detection. It has been published through three peer-reviewed research papers at top NLP conferences.
‘HateCheck: Functional Tests for Hate Speech Detection Models’ was published at ACL 2021. It was co-authored by Paul Röttger, Bertie Vidgen, Dong Nguyen, Zeerak Waseem, Helen Margetts, and Janet Pierrehumbert.
‘Hatemoji: A Test Suite and Adversarially-Generated Dataset for Benchmarking and Detecting Emoji-based Hate‘ is forthcoming at NAACL 2022. It was co-authored by Hannah Rose Kirk, Bertie Vidgen, Paul Röttger, Tristan Thrush, and Scott A. Hale.
‘Multilingual HateCheck’ is forthcoming at the Workshop on Online Abuse and Harms at NAACL 2022. It was co-authored by Paul Röttger, Haitham Seelawi, Debora Nozza, Zeerak Talat, and Bertie Vidgen.
Research Using HateCheck
HateCheck has been used in state-of-the-art research into AI for hate speech detection.
Ma et al. (2021): ‘Dynaboard: An Evaluation-As-A-Service Platform for Holistic Next-Generation Benchmarking’
Vidgen et al. (2021): ‘Learning from the Worst: Dynamically Generated Datasets to Improve Online Hate Detection’
Sen et al. (2022): ‘Counterfactually Augmented Data and Unintended Bias: The Case of Sexism and Hate Speech Detection’
Get In Touch
We’d love to hear how HateCheck is working for you.
Send us your feedback and questions!