Łukasz Szoszkiewicz
Adam Mickiewicz University | The Neurorights Foundation | l.szoszkiewicz@amu.edu.pl
I research how AI and neurotechnologies challenge human rights law, and I build practical tools for legal and human rights research.
I am an Assistant Professor at Adam Mickiewicz University, Director for European Affairs at The Neurorights Foundation, and a 2025-26 re:constitution fellow hosted by HURIDOCS.
Previously, I was a Fulbright Scholar at Columbia University’s Neurotechnology Center, where I focused on mental privacy and speech decoding.
My work sits at the intersection of international human rights law, legal informatics, and computational social science. I focus especially on the UN framework, mental privacy, digital technologies, social rights, and the problem of making complex legal materials easier to search, analyze, and use.
I currently work on neurorights and mental privacy as well as building search tools for human rights (e.g. UN Treaty Bodies documents).
What I do
- Research: I study how emerging technologies, especially AI and neurotechnologies, reshape human rights obligations and regulatory frameworks.
- Build: I develop open-source search and analysis tools for human rights case law and UN documents, including dashboards for the Polish Constitutional Court, the European Court of Human Rights, and UN Treaty Bodies materials.
- Teach: I design courses and workshops that help lawyers, researchers, and human rights professionals use computational methods and AI tools in practice.
I am especially interested in collaborations on human rights, legal tech, and AI governance.
news
| May 11, 2026 | 📖 The UN Human Rights Database now covers 4,327 jurisprudence decisions from all 8 Optional Protocol treaty bodies — 98.4% of the OHCHR JURIS catalogue, including OCR-recovered scanned PDFs from the 1980s–90s. Total corpus: 178,659 paragraphs across 4,687 documents, with 173 Special Procedures reports added in preview. |
|---|---|
| May 9, 2026 | 🔍 Launched Ask (BETA) in the UN Human Rights Database — a retrieval system that answers legal questions using verbatim Committee text rather than AI-generated summaries. It rewrites queries into legal language, retrieves candidates via hybrid search (BM25 + dense embeddings), and ranks them with a cross-encoder. Independent faithfulness evaluation: 0.91 ±0.01. No hallucinations by design — only text the Committees actually wrote. |
| May 8, 2026 | ⚖️ Major update to the ECHR Dashboard: paragraphs are now rebuilt from HUDOC source DOCX files rather than scraped HTML, giving source-exact text for every judgment. The update also adds citation extraction — each case now shows which ECtHR decisions and domestic-law sources it cites and which later cases cite it back. French-language judgments are included for the first time. |
| Mar 26, 2026 | 🧭 Participating in the EFRIS developers workshop as part of the assessment of the EU Fundamental Rights Information System for the European Union Agency for Fundamental Rights. Looking forward to discussions on interoperability, search design, and realistic AI use cases for human rights information systems. |