About MagenDetect

MagenDetect was built to solve a specific problem: the tools that exist for content moderation treat antisemitism as a subcategory of generic "hate speech" — if they detect it at all. They miss coded language, fail to distinguish reporting on antisemitism from exhibiting it, and produce no audit trail that a regulator would accept.

The EU Digital Services Act created real legal obligations for platforms to detect and mitigate systemic risks, including the spread of content that incites hatred. But it did not create the technical standards or tooling to meet those obligations. MagenDetect fills that gap.

Our detection model is grounded in the IHRA Working Definition of Antisemitism — the most widely adopted institutional framework for defining antisemitism, endorsed by 43 countries and the European Parliament. Our four-stage pipeline goes beyond keyword matching: it understands context, attribution, and the difference between a journalist condemning a trope and a publisher endorsing one.

We serve news publishers, digital platforms, and compliance teams who need to demonstrate due diligence to regulators without over-censoring legitimate journalism.