Hate speech is a widespread phenomenon in digital environments, with severe consequences for individuals and society. Counterspeech by users witnessing the incidents, the so-called bystanders, can help mitigate these effects and foster a more benign digital public sphere without limiting the freedom of expression. However, most bystanders do not intervene due to various barriers. Recent advancements in artificial intelligence (AI), particularly large language models (LLMs), offer new possibilities to support counterspeech efforts. At the same time, implementing AI for counterspeech poses immediate and long-term risks that require careful consideration. This report provides a consolidated overview by authors from different academic disciplines as well as practitioners, discussing the risks and potential of AI for counterspeech. It was developed by this international and interdisciplinary group of experts at a workshop funded by the Bavarian Research Institute for Digital Transformation (bidt) and organised at LMU Munich in February 2025.

Obermaier, M., Buerger, C., Frischlich, L., Bund, L., Carathanassis, F., Clausen, A., Doseva, S., Eissfeldt, J., Garland, J., Grimme, C., Ghazi-Zahedi, K., Haim, M., Herderich, A., Li, Y., Oetting, H., Puschmann, C., Rho, E., Stoll, A., Wenninger, A., & Ziegele, M. (2026). The Role of AI in Counterspeech: Assessing risks and potential. Bavarian Research Institute for Digital Transformation (bidt), 21. (content_copy)