A Hungarian Jewish organisation is using artificial intelligence to transform how antisemitism is monitored across Europe, aiming to create a unified, data-driven system that can identify trends, support legislation, and improve education. Its leaders argue that understanding antisemitism accurately is a prerequisite for effectively combating it, reports The Jerusalem Post.
The Action and Protection Foundation (TEV) was established in 2012 through cooperation between Hungarian Jewish communities and civil society groups and began operating in 2013. One of its core missions has been the systematic monitoring of antisemitic activity in Hungary. While many Jewish organisations rely on community reporting, TEV’s leadership believed that long-term monitoring would provide deeper insight into how antisemitism develops and how best to respond through education, policy, and law enforcement.
For nearly a decade, TEV focused on traditional media such as newspapers, radio, and television. Around three years ago, however, the organisation concluded that this approach was no longer sufficient. Much of contemporary public discourse, particularly the most extreme rhetoric, has shifted to social media platforms, which are largely unregulated and shape public sentiment far more rapidly than mainstream outlets.
TEV Chairman Kálmán Szalai, who previously worked in software development, led the creation of an AI-based monitoring system designed specifically to track antisemitic content online. The system was trained using large datasets of antisemitic material, enabling it to identify relevant keywords, phrases, and patterns. It now scans platforms including Facebook, Instagram, YouTube, LinkedIn, and Reddit.
The software collects posts and prioritises them based on engagement, such as likes, comments, and shares, as well as contextual indicators suggesting antisemitic intent. Each item is summarised, after which a human analyst currently verifies whether the content is antisemitic and categorises it. Categories include Holocaust denial, conspiracy theories, anti-Judaism, structural antisemitism, classical antisemitism, and antisemitism expressed through hostility to Israel.
Szalai expects that further automation will eventually reduce the need for human verification, enabling deeper, faster analysis of trends across countries. Importantly, the system does not classify all criticism of Israel as antisemitic. Instead, it incorporates the International Holocaust Remembrance Alliance (IHRA) working definition of antisemitism to identify when criticism crosses into demonisation or delegitimisation.
TEV’s data shows that in Hungary, online anti-Israel sentiment has become slightly more prevalent than classical antisemitism. Classical antisemitic content appears more frequently among male users, while anti-Israel narratives are more common among women. Szalai attributes this in part to the emotional impact of images from Gaza circulating online. Younger users also tend to express antisemitism through anti-Israel language rather than traditional tropes, reflecting changing online discourse rather than reduced hostility.
The system is currently operating in France, Germany, and Sweden and has no language limitations. TEV hopes to expand it across Europe, arguing that the absence of common metrics makes it difficult to compare antisemitism levels between countries or to design effective responses. Different legal frameworks and privacy rules mean the system must be adapted locally, but the core methodology remains consistent.
The data collected has already supported legislative initiatives and referrals to law enforcement in Hungary. Szalai cited a 2010 law criminalising Holocaust denial as an example of effective deterrence, noting that TEV’s September data showed only 0.01% of monitored Hungarian posts contained Holocaust denial.
Monitoring has also informed education policy. TEV has contributed to revising dozens of primary and secondary school textbooks and runs educational programmes where Jewish identity and history are discussed openly with students. According to the organisation, addressing both emotional and intellectual aspects of antisemitism is essential for long-term change.
TEV’s analysis of European Union Agency for Fundamental Rights survey data suggests significant progress in Hungary over the past decade. While Hungary ranked among the worst countries for perceived antisemitic harassment in 2013, by 2018 it had become one of the better performers. In 2013, 90% of Hungarian Jews viewed antisemitism as a significant problem; by 2023, that figure had fallen to 65%.
Szalai noted a persistent discrepancy between online and offline antisemitism across Europe. In parts of Eastern Europe, antisemitic views may be expressed more openly but result in fewer physical incidents. In Western Europe, fewer people voice such views publicly, yet violent incidents are more frequent. He suggested this may be influenced by political norms, migration patterns, and the centrality of Israel-related discourse in Western societies.
TEV argues that its data-driven approach offers a way to understand these complexities and design more effective responses. “You can only fight antisemitism if you first understand antisemitism,” Szalai said.
More News
