What the AI Act means for Swedish municipalities
The EU AI Act sets new requirements for how AI is used in the public sector. Here is what decision-makers, IT managers and heads of service need to know before procurement.
A risk-based framework
The AI Act divides AI systems into four risk levels. Services that make decisions about individuals’ rights, such as social services and education, are in focus.
The regulation does not aim to stop digitalisation, but to ensure transparency, traceability and human control.

A human decides
The most important requirement for the public sector is that a human is always in control. An AI system must never make official decisions on its own.
Interpreto is built to help prepare case material. It summarises and suggests text, but the caseworker always reviews, corrects and decides.
Data that stays in the EU
Alongside the AI Act, Sweden’s Public Access to Information and Secrecy Act (OSL), the GDPR and the Schrems II ruling all apply. Sensitive information about children, victims of crime and people receiving assistance must not be exposed to foreign laws.
Four questions to ask in procurement
Does the supplier train its models on your data? Is it operated in Swedish or European data centres? Is it a support tool, with a human making the decisions? Is every piece of material logged so that it can be traced?
More insights
Want to try it in your organisation?
Book a demo and we’ll show you how your own templates are filled in automatically.
