Principles

Neutral. Not for sale.

The database's only value is that it can be trusted. These are the rules that keep it honest — and what you, as a reader, can expect from us.

Neutrality

Status can never be bought. A data point becomes confirmed in exactly one way: publicly verifiable evidence — never through payment, partnership or pressure. The site is funded solely through donations, and donations buy nothing: no content, no status, no tone, no placement. We accept no earmarking, sell no ads, take no affiliate compensation and offer no paid "verifications". Donations from providers that appear in the comparison are disclosed openly. Any assignments or employment with compared providers are disclosed as conflicts of interest. OpenSverige is independent and has no ties to the providers being compared; all trademarks belong to their owners.

Responsibility & how you use the data

This site is an editorially independent review database that compiles publicly available information about AI infrastructure in Sweden and the EU. Every data point is reported with source, quote and date, plus an evidence state: claimed means we have only seen the provider's own statement, confirmed means the data point is supported by publicly verifiable evidence. Information may become out of date; always check against the linked source and directly with the provider before making a decision — especially in procurement or contracting. The content is a review compilation and opinion-forming, not advice, and we make no guarantees as to completeness or currency.

Corrections

Found an error? Report it in our Discord with a source. Confirmed errors are normally corrected within seven days, every correction is logged openly in the change history, and earlier versions are preserved. We never remove correct, sourced information on request — but information that concerns you as a person (for example a name in ownership details) you can always request to be corrected or removed, the same way. Providers: how a data point gets confirmed.

How we fetch data

We read only public, ungated pages, with an identifying User-Agent, respect for robots.txt and machine-readable TDM reservations (tdmrep, noai) in all fetch engines, and a reasonable fetch rate. We never store whole pages — only structured data and short quotes with a source (section 22 of the Swedish Copyright Act). The scrape replaces earlier observations; it does not accumulate the sources' material. If you object to your site being fetched: let us know in Discord and we will pause the source within seven days pending a dialogue.

License — what you may do with the data

Our dataset — the selection, the structure, the status model and the enrichment — is licensed under CC-BY-4.0: build on it freely, with attribution. Two caveats: individual facts (a price, a location) are free and need no license at all, and the evidence quotes are reproduced under the right of quotation and are not covered by our license — they belong to their authors.