Decoding Antisemitism Online: Linguistic and Multimodal Challenges in the Age of AI

Matthias Jakob Becker^1*Jordan Blatter²Oksana Stanevich³

• ¹University of Cambridge, Cambridge, United Kingdom
• ²Blue Square Alliance Against Hate, Foxborough, United States
• ³Tel Aviv University, Tel Aviv-Yafo, Israel

This article investigates the linguistic and computational challenges of detecting antisemitism in digital communication, integrating discourse-analytical and artificial intelligence (AI) perspectives. It demonstrates how antisemitic discourse spans a continuum from explicit incitement to implicit and coded expressions, whose interpretation often depends on contextual and cultural knowledge. Drawing on empirical case studies from the Decoding Antisemitism project—YouTube reactions to the Hamas terror attack of 7 October 2023 and to the antisemitic double murder in Washington, D.C., in May 2025—the analysis shows how antisemitic discourse has become normalized in mainstream digital spaces. The aftermath of 7 October was marked by open glorification of violence, whereas the Washington case centered on denial, irony, and the inversion of victimhood—together illustrating the normalization and diversification of antisemitic communication online. Building on these findings, the article discusses the methodological and computational implications for annotation and model design. It emphasizes the specific challenges antisemitism poses for automated detection—including semantic ambiguity, pragmatic drift, multimodal signaling, and data scarcity—and evaluates emerging computational approaches from transformer-based fine-tuning to retrieval-augmented and context-engineered large language models (LLMs). The study concludes that confronting digital antisemitism requires sustained collaboration between linguists, data scientists, and policymakers to develop context-sensitive, transparent, and ethically grounded AI systems capable of reliable interpretive reasoning.