It has now been little over one month since the Ethical Data Initiative (EDI) hosted its first annual multi-stakeholder town hall meeting titled “Shaping the Future of Data Ethics”. From early in the morning on Monday the 31st of March 2025, EDI affiliates, data ethics experts from around the world, as well as some of the EDI’s key stakeholders gathered at the TUM Think Tank in Munich. The inaugural townhall aimed to discuss the progress of the EDI in its first year of inception, and to organise and shape community activities for the upcoming future.

The day began with a provocative introduction from EDI Director, Professor Sabina Leonelli who set the stage for the day’s discussion. In her talk, Professor Leonelli outlined the importance of focusing on ethical data in a modern data management climate that is characterised of understanding ethics as separate from science, where the role of AI has accelerated the urgency of privacy, trust and quality concerns, misinformation has become endemic, and where the current geo-political atmosphere has become so hostile to anything deemed inclusive or diverse. To tackle these pressing challenges, Leonelli argued that we must think beyond explainability (or the idea that by simply describing technology means we will understand it and that in turn will somehow tame it); beyond personal privacy (and instead recognise the sensitivity of both social and environmental data as well); and beyond data transfer (– should any data be owned at all? Are the FAIR data principles even fair? And should we instead focus on issues of data sovereignty, misinterpretation, and extraction).

As part of this reconceptualization, and indeed for the use of data to be equitable across diverse landscapes, a key challenge for data ethics, as Leonelli argued, would be supporting capacity building for historically marginalised actors. Not only is this an imperative to counter the widening inequalities and high resource biases found across the globe, but places a strong emphasis on the situated locality for domain and challenge-specific research and tools. Moreover, it highlights the expertise, which has been grossly underestimated and unrewarded such as the work of data stewards, as well as the communities of practices needed to shape ethical data futures, including SMEs, public administration, civil society organisations, and local and indigenous experts.

Put together, this could constitute a human-centric data science (which supports fast science over rushed science and understands data work in relation to the socio-environmental needs and prospective data reuse scenarios) and participative data governance (where policy work is aligned with scientific work with a transdisciplinary emphasis particularly from those in low-resourced settings rather than big-tech). This ambition and practical aim of the EDI, Leonelli concluded, is drawn from philosophical, historical and social scientific studies of data use, together with domain specific expertise (such as statistics, machine learning, biology, biomedicine, epidemiology, public health, marine and agricultural science, humanities, social science and the arts).

Before the next session began, an important and sentimental minute of silence was called for in grace of remembering the late and great Professor Fraser Taylor’s contributions to the Data Ethics community. This touching moment offered a reminder to remember those who came before, and the tireless dedication it takes to keep conversations of ethics at the forefront of policy and academic discussions.

Following on from this, members of the EDI from the University of Exeter, Dr. Suchith Anand & Kathryn Bailey, gave an extensive overview on the work carried out over the past year concerning the EDI’s first core pillar: Policy and Networks. This included how the EDI is building alignment with international research frameworks works such as UNESCO recommendations on Open Science and the CODATA Data Ethics Policy Briefs. Moving from this some of the key policy and network stakeholder engagement were described, including participation at the G20 New Delhi conference on for Responsible AI, workshops with UKRI, participation in COARA working groups for research assessment, attendance at the Research Data Alliance working groups, as well as setting up training courses with the African Institute for Capacity Development.

Outreach and engagement updates were then offered, describing the new range of communication channels such as the website, Bluesky, newsletter, LinkedIn Group the EDI has set up. As well as the number of events set up to promote the outreach of EDI activities, including the ACU Vice Chancellor Summit, The Association of African Universities Europe Regional Office, The Ethical Data Discussion Series, The Campaign for Data Ethics in Education, and The Research Assessment for Ethics, Data and AI discussion series.

Following this, Dr. Kim Hajek & Dr. Paul Trauttmansdorff provided a summary of the fantastic work carried out over the past year on the EDI’s second core pillar: Education. This presentation covered some of the new EDI resources located on the EDI website. These included a list of readings and resources on the seven core topics – injustice and bias, inequity and access, privacy and confidentiality, openness and ownership, misuse and error, environmental damage and sustainability, and artificial intelligence – relating to why one should worry about ethical data (Hajek et al., 2025), blog posts covering a range of reports, summaries and reviews of EDI work and activities, and a data ethics glossary which has been built for a broad public audience without any prior expertise or knowledge of data science language.

On top of this, an overview of other education activities – not on the website but IRL instead – were covered. A central aspect of this was the EDI’s role in facilitating data clinics. The main principles behind “data clinics” is to provide an interactive collaboration and teaching format to bridge theory and practice. This concept works by bringing together invited participants who present a real-world challenge based on their actual data practices. Students from TUM are then asked to collaborate in order to develop insights from multiple perspectives which can be used as practical advice to navigate questions around responsible data practices. Data clinics, as Dr Hajek & Dr Trauttmansdorff describe, are set with the goal to empower participants with actionable insights to handle complex data governance and ethics issues like privacy, discrimination, transparency, openness, and accountability, and foster wider cultures of ethical data work. The first data clinic, will be kickstarting this May with the question of how can one foster transparency and explainability in the use of AI within the African fintech ecosystem to ensure fairness, accountability, and user trust? This session is in collaboration with the Center for Law & Innovation, Certa Foundation in Kigali, Rwanda.

Additionally, education material from the EDI is being planned to be uploaded to LabXchange – a content library with thousands of free educational resources spanning teaching guides, annotated videos and presentations. LabXchange will be the perfect home for EDI resources given its global coverage and its focus on serving underrepresented communities with an infrastructure which supports low-bandwidth access. As part of this, the EDI will set up a Data Ethics Cluster which will bring together structured and modular learning programmes from EDI resources and activities, as well as videos, quizzes, games and lecture materials. With the plan of eventually being available in a multi-lingual format. LabXchange will also house the planned data stories – short videos exploring a particular case study in data ethics and its implications. These stories will come from real academic research and involve interactive elements such as pop-up explanations and a choose your own adventure branching structure. The first planned story will come from myself (Nathanael Sheehan), covering my research on the implications of mandating file formats for data sharing and the participatory exclusion this entails particularly for those working in low-resourced environments.

The last EDI core pilar: research, was then presented by Dr. Silvia Milano and Dr. Stephan Güttinger. Both researchers adopt a cross-disciplinary approach to studying data which involved the integration of philosophy with history, sociology, and Science and Technology Studies (STS), and builds on close interaction with data practitioners. Two key strands of research being undertaken were described respectively as the role of recommender systems in science, as well as the ethical and societal challenges these systems produce. And the role of datafication in the laboratory sciences, paying particular attention to the dataveillance of research and the implications of digitals tools being superior to analogue spaces.  

After a well needed coffee break, the town-hall reconvened to hear a series of flash talks and keynotes by EDI affiliates. This began with an eye-opening flash talk by Dr. Denis Newman-Griffis, where they introduced the GRAIL project and its research on the shared practices on responsible AI in research funding. Following on from this, Dr. Ed Humpherson from the Office for Statistics Regulation (OSR) in the United Kingdom gave a compelling talk on the role of communicating data in the public sphere, paying attention to the ways data becomes politicised and the role the OSR take to combatting this (for more on this see Dr. Ed Humpherson’s blog on this talk). Wrapping up the final flash talk was Dr. Francis P. Crawley from the International Data Policy Committee & CODATA-ISC. Francis charmed the audience with his reflections on the state of ethics in today’s digital landscape, and through his own reflections on the work of Socrates and Plato, offered the audience the opportunity to imagine a world without data –or if such a world is even possible?

The final session before lunch was delivered by Professor Xiao-Li Meng (Editor-in-Chief at the Harvard Data Science Review) who gave a keynote on Privacy, Data Privacy, and Differential Privacy. While the first of these two types of privacy are quite well known, Differential Privacy is far less known but refers to a probabilistic and statistical framework for regulating procedures with the aim of protecting data privacy by injecting noise while maintaining some data utility. By drawing historical parallels from the “the right to be alone” written by Samuel D. Warren II and Louis Brandeis in 1890 and the “the right to be forgotten” part of the General Data Protection Regulation (GDPR) written over 125 years later, Professor Xiao-Li Meng’s talk illustrated the long need for new and different kinds of privacy to be written into law for society. After stating this thesis, he brought the audience back to statistics 101 and in his both entertaining and stimulating style, explained Data Differentiability in a way that even the least technical people could understand.

After lunch, the second session of flash talks commenced with an initial presentation by Dr. Louise Bezuidenhout and her insights from the UNESCO Chair in Diversity and Inclusion in Global Science. One key point here was the introduction of “The Mulitversatory” an initiative aimed to promote pluralistic methods for research assessment using a decentralised and federated research information ecosystem. Dr. Davide Posillipo and Dr. Giuseppe Primiero then presented work on the MIRAI toolbox. The toolbox, as they described is an All-In-One tool for Bias, Risk and Opacity of AI systems and was born out of collaboration between academic and industrial partnership. The final flash talk of the day was given by Dr Magdalena Eitenberger from the University of Vienna. Dr Eitenberger’s presentation highlighted some of the latest developments of the Data Solidarity project, which included the public value assessment (PLUTO) tool, the data solidarity glossary and updates on a white paper published on the topic.

Moving from talks to community participation, the afternoon then focused on attendees going into break out groups based on the three core pillars of the EDI (Policy & Networks, Education, Research), to consult on how the EDI could develop some of the community feedback into the next stages of development. Break out groups were headed by EDI staff members with a self-selected community member acting as the rapporteur to present the discussion back to the townhall. Community members found gaps, possibilities, and connections for each of the EDI pillars which were all taken into consideration for the EDI’s next stages of planning.

After the final coffee break of the day, a final keynote was delivered by Hilary Hanahoe the Secretary General of the Research Data Alliance (RDA). For those unfamiliar with the RDA, they are a global community-driven organisation with the vision of bringing together researchers and innovators to openly share data across technologies, disciplines and countries to address the grand challenges of society. Guiding this vision are six core principles: Openness, Consensus, Inclusivity, Harmonisation, Community-Driven and Non-profit & Technology-neutral. Since its launch in 2013, the RDA community has grown, from an original membership of 250 to now over 15,500 (myself being one of them), from 14 working groups to now over 97 working groups and over 200 groups since its inception, and from 0 outputs and recommendations to now over 100. The RDA has been instrumental in advocating, establishing and educating data practitioners in some of the most influential data management frameworks used in science and industry today. This includes a fierce endorsement of both the FAIR data principles and the TRUST principles for digital repositories, let alone an RDA group setting up the CARE Principles for Indigenous Data Governance. Hilary’s talk finished with a call to action, for not only those in the room to join and shape RDA activities, but for the RDA and the EDI to establish a strong alliance and working relationship, so the RDA, in Hilary’s enthralling words, could drink its own champagne.

The townhall was concluded with some final remarks by Professor Sabina Leonelli, who summarised the town hall as being a generative and fruitful day of community building and idea generation. Although by this point in the day, many were tired, there was a shared feeling that the EDI were on the right path, not only in the work already being carried out, but more importantly in creating and bringing together a community who could adopt a collective life of their own, creating a self-sustaining force to carry out the vision and practices envisioned by the EDI to the present and beyond.

**

Presentations and the programme for the EDI townhall meeting are available on zenodo: https://zenodo.org/records/15174049

References:

K. M. Hajek, P. Trauttmansdorff, S. Leonelli, S. Guttinger and S. Milano, “How to Foster Responsible and Resilient Data: The Ethical Data Initiative,” in Computer, vol. 58, no. 4, pp. 95-99, April 2025, doi: 10.1109/MC.2024.3522696.