Automating Society Report 2020

Automating Society Report 2020 is a 297-page multi-country mapping of automated decision-making systems across Europe, jointly published by AlgorithmWatch and the Bertelsmann Stiftung on 28 October 2020 and edited by AlgorithmWatch's Fabio Chiusi as project manager and co-editor under the general editorial direction of Matthias Spielkamp, with country chapters researched by a network of civil-society and journalist contributors. It is AlgorithmWatch's flagship recurring mapping product — the second edition in the series, expanded from the 12-country 2019 edition to 16 countries (Belgium, Denmark, Estonia, Finland, France, Germany, Greece, Italy, Netherlands, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom) plus an EU-level chapter, and was placed on the Bertelsmann Stiftung side inside its reframe[Tech] – Algorithms for the Common Good programme, free to download under a Creative Commons Attribution 4.0 International licence.

The report's central argument — captured in Fabio Chiusi's framing line "We're no longer just automating society. We have automated it already" — is that automated decision-making has moved decisively from experimental pilots into mainstream public-sector and private-sector deployment across Europe, while the underlying systems remain opaque, poorly audited, and largely outside democratic debate. Each country chapter pairs a journalistic narrative "story" with a research section cataloguing concrete ADM systems then in use; together the chapters document over 100 specific cases of automated decision-making across welfare, education, policing, justice, healthcare, and consumer-credit contexts. The two findings the report itself flags most heavily are the rapid spread of face-recognition systems — described as "nearly absent from the 2019 edition" but trialled or deployed in schools, airports, stadiums and policing across the surveyed countries by 2020 — and the persistence of an opacity gap in which existing national algorithmic-transparency obligations (France's and Spain's are named) are not being enforced in practice. The report concludes with a three-pillar Policy Recommendations section addressed primarily to EU institutions: legislate to require public registers of ADM systems in the public sector and statutory data-access frameworks for independent researchers; create a meaningful accountability architecture for ADM (multi-stakeholder auditing, civil-society watchdogs as a recognised accountability mechanism, and an EU-level ban on face-recognition deployments amounting to mass surveillance until adequate safeguards exist); and invest in algorithmic literacy through independent national centres of expertise and inclusive democratic debate. The recommendations slot directly into the policy positions AlgorithmWatch was then taking forward into the European Commission's 2021 AI Act proposal.

The report was launched on 28 October 2020 in a 13:15–14:30 CET online event anchored by Sarah Fischer for the Bertelsmann Stiftung side and Matthias Spielkamp and Fabio Chiusi for AlgorithmWatch, with a panel featuring MEPs Miapetra Kumpula-Natri (S&D) and Sergey Lagodinsky (Greens/EFA), European Commission AI Team Leader Irina Orssich, Microsoft's Cornelia Kutterer, and Hertie School ethics-and-technology professor Joanna Bryson — a line-up that itself maps the institutional audience the report was pitched at (European Parliament, Commission, large platform companies, and academic ethics). The mapping work was continued in a Swiss Edition published on 28 January 2021 extending the methodology to Switzerland, and the report's pattern of pairing per-country ADM cataloguing with EU-level policy asks has since been used as the reference baseline by European civil-society researchers and journalists working in the AI Act run-up.

Within the corpus, Automating Society Report 2020 is the first Continental European publication on the slate — closing the geographic gap left by the prior Publications, which had clustered in the United States (Unmasking AI, Comply To Fly?, Bug Bounties for Algorithmic Harms, A Hazard to Human Rights, Losing Humanity) and the United Kingdom (Participatory Data Stewardship, Not a drop to drink). It is also the corpus's first multi-country systematic-mapping report — a publication-shape distinct from the auditing artefacts (the AJL papers), the legal-academic foundational texts (Losing Humanity, A Hazard to Human Rights), the framework-building report (Participatory Data Stewardship), the campaigning evidence base (Not a drop to drink), and the long-form policy plan (A Narrow Path). Where those earlier publications each anchor a single argument or programme, Automating Society Report 2020 anchors a continuing inventory: its function in AlgorithmWatch's own working practice is to keep a running European-wide record of which ADM systems are actually being deployed, on what populations, and under what regulatory cover — the empirical floor on which the organisation's subsequent EU AI Act advocacy is built.