AlgorithmWatch

AlgorithmWatch is a Berlin-headquartered non-profit research-advocacy organisation working on the social, legal, and ethical implications of automated decision-making (ADM) systems in Germany, Switzerland, and across the European Union. Its self-description frames the organisation as a watchdog whose function is to ensure that automated and AI-driven systems used by governments and large companies "strengthen justice, democracy, human rights and sustainability, rather than weaken them". The organisation runs an unusual programme mix for the European digital-rights field — original policy advocacy, in-house investigative tech journalism, and large-scale participatory "data donation" research projects that recruit members of the public to collect evidence on how platform algorithms behave — and sits at the centre of the European civil-society coalition that pressed for fundamental-rights protections in the EU AI Act.

Founding and structure

AlgorithmWatch was founded as a non-profit gGmbH in 2017 in Berlin (the charitable-status determination was issued on 14 July 2017), building on a 2016 public-launch announcement at the re:publica conference. The originating founders included journalist and digital-rights researcher Matthias Spielkamp, data journalist Lorenz Matzat, computer scientist Katharina Anna Zweig, and ethicist Lorena Jaume-Palasí; Spielkamp and Matzat remain the equal-partner shareholders of the gGmbH and Spielkamp serves as Executive Director. The organisation is structured as a non-profit limited liability company (gemeinnützige Gesellschaft mit beschränkter Haftung) governed by a supervisory board and shareholders, with a fellowship programme for outside researchers and practitioners. By 2022 it reported €1.1 million in grant revenue and a staff of roughly 25 across the German and Swiss entities.

The senior leadership layer that has emerged is a function-area structure: Matthias Spielkamp as Executive Director, Dr. Angela Müller as Executive Board Member of AlgorithmWatch and Executive Director of AlgorithmWatch CH (the Swiss sister organisation, see § AlgorithmWatch Switzerland below), Kristina Hübner as Deputy Director, Kilian Vieth-Ditlmann as Head of Policy, Dr. Oliver Marsh as Head of Tech Research, Dr. Nicolas Kayser-Bril as Head of Journalism, and Nina Galla as Head of Communications & Campaigns. Spielkamp's external roles are themselves a useful map of the German civil-society digital-rights field: he serves on the governing board of the German section of Reporters Without Borders and of Stiftung Warentest, holds an Expert Committee role at Germany's UNESCO Commission, sat on the Global Partnership on AI from 2020 to 2022, and was appointed to the German Digital Services Coordinator advisory council for the 2024–2026 term.

Programme areas

The organisation's published programme structure is built around four interlocking lines of work. The first is policy advocacy on European, German, and Swiss regulation of AI and automated systems — directed at the EU AI Act, the Digital Services Act, the Council of Europe AI Convention, and national-level workplace and public-sector ADM rules. The second is tech research that audits specific algorithmic systems through reverse-engineering, scraping, and platform-behaviour experiments. The third is investigative tech journalism — a dedicated newsroom that produces reporting on algorithmic harms and platform power, an unusual organisational shape for a civil-society advocacy non-profit. The fourth is participatory "data donation" research in which volunteers from the general public install tools or hand over downloaded platform data to enable independent research on otherwise-opaque algorithmic systems. Across all four, the unifying theory of change is that empirical evidence about how specific ADM systems actually behave — gathered, ideally, with the participation of the people those systems are applied to — is the precondition for democratic accountability over them.

Signature projects

Three projects mark the public profile of AlgorithmWatch's participatory-research line and are central to why the organisation matters in the corpus's terms.

OpenSCHUFA (February–November 2018) was the first widely-noticed European data-donation campaign on an algorithmic-accountability target. Run jointly with the Open Knowledge Foundation Germany, the project asked German residents to use their statutory right to a free annual credit-record disclosure from SCHUFA — the dominant German credit-scoring bureau, whose scores effectively gate rental tenancies, mobile-phone contracts, and consumer credit for some seventy million adults — and donate the resulting records to a public dataset for analysis. The campaign ultimately generated more than 4,000 donated SCHUFA records, prompted more than 30,000 subject-access requests to SCHUFA, and produced — in partnership with data journalists at Bayerischer Rundfunk and Spiegel Online — findings published in November 2018 showing systematic anomalies and GDPR-compliance failures in SCHUFA's data handling, even as the underlying scoring algorithm itself remained legally protected as a trade secret. OpenSCHUFA established the participatory-data-donation pattern that the organisation has used since.

Instagram newsfeed monitoring (March 2020–July 2021) scaled that pattern to a Meta-owned platform. A browser add-on installed by some 1,500 volunteers collected the Instagram newsfeed each volunteer saw, enabling AlgorithmWatch to publish findings that the recommendation algorithm preferred posts showing particular body presentations and dampened the reach of textual political content. In May 2021 — shortly after Facebook had shut down NYU's Ad Observatory project on similar grounds — Facebook approached AlgorithmWatch alleging that the project breached Section 3.2.3 of its Terms of Service (the prohibition on automated data collection) and, separately, the GDPR rights of non-consenting users whose content had appeared in volunteers' feeds. AlgorithmWatch terminated the project and deleted the collected data on 13 July 2021 and went public with the dispute on the grounds that "an organisation the size of AlgorithmWatch cannot risk going to court against a company valued at one trillion dollars". The episode became one of the most-cited European cases for the argument that statutory researcher-access provisions — eventually embedded in the Digital Services Act — were needed because terms-of-service-based access could not be relied on.

DataSkop (2020–ongoing) is AlgorithmWatch's continuing data-donation infrastructure. It is run as a five-organisation consortium with Europa-Universität Viadrina, Fachhochschule Potsdam, Universität Paderborn, and Mediale Pfade, funded by the Federal Ministry of Education and Research (BMBF) under the "Mensch-Technik-Interaktion für digitale Souveränität" (human–technology interaction for digital sovereignty) programme. The methodology shifted, after the Facebook episode, away from browser-scraping toward asking users to invoke their GDPR-mandated platform-data-export rights and upload the resulting archives — a legally safer participatory architecture. DataSkop has produced cycles on YouTube recommendations during the 2021 German federal election and on TikTok (data collection January–April 2023), the latter yielding insufficient data for reliable conclusions, a finding the project's transparency about itself treats as part of the public record.

Alongside these participatory projects, AlgorithmWatch produced the multi-year Automating Society reports — the 2020 edition (28 October 2020, in partnership with the Bertelsmann Stiftung) mapped ADM systems across sixteen European countries and EU institutions, and has been used as a reference baseline by European policy researchers since.

EU AI Act advocacy

AlgorithmWatch is one of the small number of organisations that worked the European Commission, Parliament, and Council tracks on the EU AI Act from the 2021 proposal through the 2024 adoption. Its core positions — established in an August 2021 consultation response — were that the Act should comprehensively prohibit biometric recognition systems capable of enabling mass surveillance in public space, replace the proposal's technology-list approach with an impact-based regulatory architecture, require mandatory fundamental-rights impact assessments for high-risk systems, give workers subject to ADM systems statutory information rights, and create legally binding data-access frameworks for public-interest research. The organisation co-drafted and co-signed the November 2021 joint civil-society statement titled "An EU Artificial Intelligence Act for Fundamental Rights", signed by 115 organisations including European Digital Rights (EDRi) and Access Now, and remained inside that EDRi-coordinated coalition throughout the trilogue negotiations. On the Act-text outcomes its public framing has been mixed — the partial biometric-recognition restrictions and the fundamental-rights-impact-assessment provisions are described as wins; the law-enforcement and migration-context carve-outs as a substantial loss for which the coalition continues to press through implementation guidance.

AlgorithmWatch Switzerland

AlgorithmWatch Switzerland was launched on 20 November 2020 in Zurich as a legally independent Swiss organisation under founding managing director Dr. Anna Mätzener (Angela Müller subsequently took on the Executive Director role), with seed funding from the Engagement Migros development fund. Its flagship product is the Atlas of Automation, a public database of algorithmic systems in use in Switzerland, and it is supported by the Mercator Foundation Switzerland, the Christoph Merian Foundation, the Ernst Göhner Foundation, and the Hasler Foundation (for the RAISD democracy-and-AI project). The Swiss entity is run as a sister organisation rather than a branch — its own legal personality, its own funder base, and its own programme, sharing brand and methodology with the Berlin organisation but not a single organisational entity.

Posture in the movement

AlgorithmWatch's distinctive contribution to the broader make-AI-good movement is the deliberate fusion of three practices that civil-society organisations usually do separately: regulatory advocacy, investigative journalism, and participatory empirical research run with non-specialist publics. The participatory line — recruiting tenants, Instagram users, voters, and ordinary platform users to contribute the data needed to audit specific systems — is the corpus's clearest European example of the working principle that engages non-AI publics in shaping how AI is built and held accountable. Its EU-track advocacy is the strongest single source of fundamental-rights framings that ended up in the AI Act text. And its Berlin–Zurich-anchored geography — combined with its leadership of multi-country mapping projects such as Automating Society — gives the corpus its first organisational coverage of the German-speaking and continental-European digital-rights field, which had been a substantial geographic gap before this entry.