Compliance

Impactful algorithm

An algorithm that materially affects citizens, services, or rights.

How the laws define this

Both the National Algorithm Register and the Amsterdam Algorithm Register agree that an algorithm qualifies as impactful when its operation materially affects individuals, citizens, legal rights, or the delivery of public or municipal services, thereby requiring formal registration. The primary divergence lies in how each framework categorizes the technology and defines the resulting compliance obligations. The National Algorithm Register explicitly spans deterministic systems, predictive models, scoring tools, and AI systems with significant societal or individual effects, making the main mandate the publication of structured metadata. By contrast, the Amsterdam Algorithm Register compares these systems to high-risk AI under emerging EU rules and mandates a broader operational process, prioritizing mandatory risk assessment alongside the documentation of purpose, data use, identified risks, and mitigation measures.

Synthesised from the 2 statutory definitions below. Each one is quoted in full, with its source.

How jurisdictions differ

What is distinctive in each definition, binding law in force first.

Jurisdiction	Instrument	Status	What is distinctive
Netherlands	National Algorithm Register	In Force	This definition adds that impactful algorithms include deterministic systems, predictive models, scoring tools and AI systems, and identifies them as warranting publication of structured metadata.
Amsterdam, Netherlands	Amsterdam Algorithm Register	In Force	This definition adds prioritization for risk assessment alongside documentation of purpose, data use, identified risks, and mitigation measures, and compares to high-risk AI under emerging EU rules.

What it triggers

In the Netherlands National Algorithm Register, public bodies are guided to document impactful algorithms by publishing structured metadata.source
In the Amsterdam Algorithm Register, impactful algorithms warrant registration, risk assessment, and documentation of purpose, data use, identified risks, and mitigation measures.source

Definitions (2)

An algorithm identified by the register as materially affecting citizens, public services, or legal rights and thus requiring publication of structured metadata; this includes deterministic systems, predictive models, scoring tools and AI systems that have significant societal or individual effects.

National Algorithm Register (Algoritmeregister Nederlandse Overheid)Definition 1 of 2

An algorithmic system whose operation materially affects individuals or the delivery of municipal services and therefore warrants registration and risk assessment; this category includes systems comparable to 'high-risk' AI under emerging EU rules. Impactful Algorithms are prioritized for documentation of purpose, data use, identified risks, and mitigation measures in the register.

Amsterdam Algorithm Register (Algoritmeregister Amsterdam)Definition 2 of 2

Not to be confused with

High-risk algorithm

High-risk algorithm focuses on significant impact on rights, duties, entitlements or interests, while impactful algorithm emphasizes material effects on citizens, services, or rights.

Rights-impacting AI

Rights-impacting AI specifies significant legal or material effects on people from AI outputs, whereas impactful algorithm covers broader material effects including on services.

Questions and answers

What must be done with impactful algorithms in the Netherlands?

The National Algorithm Register guides Dutch public bodies in documenting impactful algorithms by publishing structured metadata. This policy sets out public transparency standards.

What triggers registration in Amsterdam's Algorithm Register?

Algorithmic systems materially affecting individuals or municipal services warrant registration and risk assessment. They are prioritized for documentation of purpose, data use, risks, and mitigations.

How do the Dutch and Amsterdam definitions of impactful algorithm differ?

The national definition emphasizes publication of structured metadata for various algorithm types with societal effects. The Amsterdam definition adds risk assessment and documentation details, likening to high-risk EU AI.