VW Change! Fellowship
Fellowship support for research designed to shape societal change through transdisciplinary collaboration.
Responsible AI, public-interest technology, and civic systems
Orestis Papakyriakopoulos
I am Assistant Professor of Societal Computing at the Technical University of Munich and head of the Civic Machines Lab. My work studies how AI systems, platforms, and socio-technical infrastructures can become more fair, safe, democratic, and useful.
Across academia, industry, and policy engagement, I work on AI alignment, dataset governance, algorithmic fairness, political communication, and the translation of regulation into technical practice.
About
What my team works on
My research sits at the intersection of computer science, social theory, and governance. I study AI as a socio-technical system: from data collection and benchmark design to model behavior, deployment, regulation, and real-world use.
Before joining TUM in 2024, I was an AI Ethics Research Scientist at Sony AI, a postdoctoral researcher at Princeton University, and a visiting researcher at the MIT Media Lab. Earlier, I worked at TUM on political data science, platforms, recommender systems, NLP bias, and the online dynamics of democratic communication.
Civic machines are the long-term thread across this work: systems designed and governed so that technology supports equality, justice, freedom, and social inclusion rather than undermining them.
Current focus
- Pluralistic alignment and the safety of large language model agents
- Fairness, consent, and curation in multimodal and speech datasets
- How the EU AI Act and related policy frameworks translate into technical practice
- AI adoption in mission-driven organizations, civil society, and public institutions
- Platforms, political communication, transparency, and democratic accountability
Selected publications
Recent work across AI safety, fairness, platforms, and governance
- 2026
- 2025
-
2025
Safety devolution in AI agents
NeurIPS 2025
-
2025
Participatory AI and the EU AI Act
AIES 2025
- 2024
- 2024
- 2023
- 2023
Policy and public impact
Research translated into governance, standards, and public debate
Policy contributions
- Organized the Embodied AI event at the Civic Machines Lab at the TUM Think Tank, 2026
- Expert recommendations for the future of the semiconductor supply chain, Civic Machines Lab discussion paper, 2025
- AI adoption across mission-driven organizations, TUM Think Tank output, 2025
- Participated in Partnership on AI working groups that informed white papers on ethical data collection, 2024
- Research cited in the International AI Safety Report 2026
- Cited by the European Union Agency for Fundamental Rights for work on bias in NLP systems, 2022
- Cited by the U.S. Federal Trade Commission report to Congress for work on content moderation and online harms, 2022
- Cited by the European Commission and the German Parliament on political microtargeting and election technologies, 2021
- Presented work on political ads to the Public Disclosure Commission of the State of Washington, 2022
- Presented work on recommendation bias to the German Parliament study commission on artificial intelligence, 2019
Public engagement
- CNN Greece on ACM FAccT and the next phase of responsible AI, 2025
- El Periódico on the ethics of computer vision, 2025
- Times Higher Education on research and public engagement at TUM, 2024
- Financial Times on online platforms and transparency, 2021
Awards and leadership
Recent milestones
Bavaria-Queensland Research Alliance Grant
Grant support for joint work on AI and journalistic values.
Humboldt Philipp Schwartz Initiative
Grant support for research and academic capacity building.
Best Paper Award, ICML
Awarded for work on dataset diversity and responsible evaluation.
Best Paper Award, Online Social Networks and Media
Recognition for research on digital platforms and civic communication.
General Chair, ACM FAccT
Leadership in one of the central venues for fairness, accountability, and transparency.
Diversity and Inclusion Chair, ICWSM
Field-building work focused on community standards and participation.
News
Recent activity, publications, talks, and appointments
Add new items at the top to keep this page current.
General Chair of IEEE Secure and Trustworthy Machine Learning 2026
Current community leadership role alongside ongoing work on alignment, AI safety, and governance.
AAAI paper on pluralistic values in LLM alignment
New work on the trade-offs between safety, inclusivity, and model behavior in aligned language models.
Nature publication on fair human-centric image benchmark datasets
Consent-driven fairness evaluation for AI systems, building on work started at Sony AI.
TUM Think Tank output on AI adoption across mission-driven organizations
New work on how civil society and public-interest institutions are engaging with AI in practice.
Started at TUM as Assistant Professor and launched the Civic Machines Lab
Established the lab at the TUM Think Tank and expanded work on responsible AI, fairness, and policy.
Best Paper Award at ICML
Recognition for recent work on how dataset diversity should be measured rather than merely asserted.
Work presented at FAccT, CHI, and WSDM
Recent publications on speech datasets, political discourse on Reddit, and ethical speech recognition datasets.