AI models are struggling to identify hate speech, study finds (The Independent)
In their classification of hate speech across 1.3 million synthetic statements referencing 125 demographic groups, top AI content moderation systems,…
OpenAI, DeepSeek, and Google vary widely in identifying hate speech (Penn Today)
Researchers at the Annenberg School for Communication found that seven big language models, including those from OpenAI, DeepSeek, Google, and…
A Taxonomy of Response Strategies to Toxic Online Content: Evaluating the Evidence (arXiv)
In order to analyze the diverse range of counterspeech and counternarrative responses used by people, organizations, and language models, the…
New on counter-terrorism.org, 8 September 2025 (Policyinstitute.net)
This collection of recent research and initiatives offers a multidimensional view of radicalisation and violent extremism, spanning psychological, educational, technological,…
Two Weeks in Soft Security: Free Resources on Countering Extremism, Hate, and Disinformation, August 2025 (II/II)
Browse and read from a list of the 160 most popular LinkedIn resources dealing with soft counter-extremism, countering hate speech,…