METHODOLOGY
Legal & Definitional Standards
HateCheck's classification methodology is grounded in internationally recognized definitions and Dutch law.
WHY STANDARDS MATTER
Without Clear Standards, Three Things Go Wrong
Real threats are missed
Coded language, dog whistles, and semantic drift evade keyword filters. Without structured frameworks, dangerous content slips through.
Legitimate speech is mislabelled
Sharp political critique is not the same as antisemitism. Over-classification erodes trust and chills academic freedom.
Actions don’t hold up
Reports built on weak analysis fail in courts, boardrooms, and public discourse. Defensibility requires rigorous methodology.
HateCheck does not rely on a single definition of antisemitism or hate speech. Each of the primary international frameworks has strengths and limitations. Rather than choosing one and discarding others, our methodology applies all applicable frameworks in a layered approach - surfacing the most defensible classification for each piece of content.
For Dutch-specific cases, we apply the Dutch Penal Code frameworks on top of the international standards to assess criminal liability thresholds. This layered approach produces Tier 4 and Tier 5 classifications for content that not only meets antisemitism definitions but may also constitute criminal speech under Dutch law.
INTERNATIONAL STANDARD
IHRA Working Definition
International Holocaust Remembrance Alliance - Adopted 2016. Endorsed by 35+ countries including the Netherlands.
Official SourceCore Definition
“Antisemitism is a certain perception of Jews, which may be expressed as hatred toward Jews. Rhetorical and physical manifestations of antisemitism are directed toward Jewish or non-Jewish individuals and/or their property, toward Jewish community institutions and religious facilities.”
Contemporary Examples (11 Total)
ROLE IN HATECHECK
IHRA serves as the primary framework for all cases not directly related to Israel. Its broad adoption makes it the most institutionally recognized baseline for classification.
How HateCheck Uses This
Provides the primary test for whether content is antisemitic at all. Reasoning logs reference IHRA-style patterns (demonisation, dehumanisation) for Tier 3–5 classifications.
CRITICISM VS. ANTISEMITISM
The Nexus Document
Nexus Task Force - 2021. Focuses specifically on Israel-Palestine discourse and the antisemitism boundary.
Official SourceThe Two-Part Nexus Test
Speech about Israel is antisemitic when it:
Explicitly Protected Speech
The Nexus Document explicitly states that criticism of Israeli government policies, support for Palestinian rights, and opposition to occupation are not antisemitic - even when strongly stated. The “because it is Jewish” test separates political discourse from hatred.
ROLE IN HATECHECK
Nexus analysis is applied specifically to cases where the Israel-antisemitism boundary is contested, providing a structured test for distinguishing political criticism from antisemitic speech.
How HateCheck Uses This
When an incident involves Israel/Palestine, Zionism or boycotts, HateCheck asks: Is the content relying on classic antisemitic myths? Is it blaming Jews as Jews for actions of Israel? Is Israel being singled out by standards not applied to any other state?
ACADEMIC FRAMEWORK
Jerusalem Declaration on Antisemitism
Jerusalem Declaration - 2021. Developed by 200+ international scholars. Provides academic rigor and Israel-specific nuance.
Official SourceWhat is Antisemitic (JDA)
- Calling for harm to Jews or treating them as less than human
- Conspiracy theories (Jewish control of media, finance, government)
- Denying or trivializing the Holocaust
- Discriminatory double standards applied to Jews
- Attributing Israeli conduct to Jewish identity globally
What is Not Antisemitic (JDA)
- Criticism of Israeli policies comparable to criticism of other states
- Holding that Israel's founding involved ethnic displacement
- Boycott, divestment, sanctions (BDS) advocacy
- Comparing Israeli policies to other historical examples (context-dependent)
ROLE IN HATECHECK
JDA guidance is applied as a secondary check for Israel-related content, providing more granular boundaries for grey-zone cases that IHRA alone cannot resolve.
How HateCheck Uses This
Useful for borderline cases in campus debates or academic speech. Helps avoid over-classification when criticism is sharp but not antisemitic.
LEGAL FRAMEWORK
Dutch Penal Code
The Netherlands' criminal provisions governing hate speech are concentrated in Articles 137c-e of the Wetboek van Strafrecht (WvSr). These are the legal thresholds that determine Tier 4 and 5 classifications.
ART. 137C
Group Defamation (Groepsbelediging)
Criminalizes deliberately insulting a group of persons based on race, religion or belief, sexual orientation, or disability. Maximum: 1 year imprisonment or Category 3 fine.
- Deliberate (opzettelijk) intention required
- Must be made publicly
- Group defined by protected characteristic
- Insult threshold - not all offensive speech qualifies
ART. 137D
Incitement (Aanzetten tot Haat)
Criminalizes public incitement to hatred against, discrimination of, or violence against persons based on race, religion, sexual orientation, or disability. Maximum: 2 years imprisonment or Category 4 fine.
- Higher threshold than 137c
- Must be likely to provoke hostile action
- Context and likely effect are critical
- Most relevant for antisemitic incitement online
ART. 137E
Distribution of Hate Materials
Addresses distribution, public exhibition, or possession for distribution of materials containing group defamation (137c) or incitement (137d). Critical for social media sharing.
- Covers sharing/reposting - not just creation
- Hosting hateful content may trigger liability
- Platform-level obligations under DSA intersect here
- Intent to distribute required
ROLE IN HATECHECK
Dutch law is applied on top of international standards to assess criminal liability thresholds, producing Tier 4 and Tier 5 classifications for content that may constitute criminal speech.
How HateCheck Uses This
Maps verified incidents to the specific statutory requirements of Articles 137c, 137d, and 137e. Generates structured evidentiary dossiers for the Public Prosecution Service.
INTEGRATED METHODOLOGY
How They Work Together
The layered methodology integrates all frameworks at the appropriate analytical stage.
Layer 1: International Definitions
IHRA, JDA, NEXUS
Applied first to determine whether content meets any antisemitism threshold.
Layer 2: Context Analysis
Satire Check, Intent Signals, Audience Context
Contextual modifiers that can affect tier assignment in borderline cases.
Layer 3: Dutch Legal Assessment
Art. 137c, Art. 137d, Art. 137e
Applied to determine criminal law applicability for Tier 4/5 assignments.
Layer 4: Tier Assignment
Tier 1, Tier 2, Tier 3, Tier 4, Tier 5
Final classification produced with full reasoning chain.
GOVERNANCE
Safety, Privacy & Governance
No automatic external actions
HateCheck never sends emails or forms to authorities automatically. AI drafts text; humans control the send button.
Careful legal language
We say ‘Potentially criminal’, never ‘This is a crime’. The app is a reference tool, not a court.
Data isolation
Every organisation’s data is siloed. No organisation can see another’s incidents. No tracking or analytics.
See the Standards in Practice
Review our anonymized case studies to see how these frameworks are applied.
View case studies