Yonah Welker
Boards now own AI decisions that used to sit two layers below them. The EU AI Act, the OECD framework, and UNESCO’s ethics recommendation increasingly govern the same call, and they do not always agree. The hardest cases now involve AI acting in the physical world and in public services. That is where the rules are least settled, and where a wrong answer is hardest to defend.
Yonah Welker is a public technologist who works inside the EU and international bodies that write AI rules, helping organisations build defensible governance for high-stakes systems, from accessibility and designated groups to AI in physical and public deployment.
Full Profile
Why organisations work with Yonah Welker
- Direct line into the European policy machinery shaping AI rules, through working group contributions to the EU AI Code of Practice, DG CNECT’s digital and critical infrastructure work, and EU ICT and standardisation programmes.
- Published, citable expertise on how AI systems treat disability, neurodiversity, and other designated groups, with frameworks and a MOOC catalogued by OECD.AI, a specialism few governance advisers can match.
- Cross-institutional fluency across OECD, UN, UNESCO, and WHO governance processes, useful for organisations whose AI strategy now has to answer to several overlapping frameworks at once.
- Practical command of how governance applies to AI in physical and public deployment, including robotics and assistive systems, where compliance obligations under the EU AI Act and emerging harmonised standards are least settled.
- A practitioner-policy bridge rather than a pure academic, with hands-on evaluator and rapporteur roles across EU Horizon and EIT/EIC programmes, which is the lane most boards and compliance leads now need filled.
Biography highlights
- AI expert listed in the OECD.AI community directory; visiting lecturer and evaluator affiliated with MIT collaborative work, and contributor to OECD.AI’s Wonk blog.
- Contributor to the EU AI Code of Practice (working Groups 2 and 3), DG CNECT digital and critical infrastructure programmes, and EU ICT standardisation work.
- Contributor of input to OECD (G7 Hiroshima AI Process), UN (ITU and IGF), UNESCO, NIST (Risk Management Framework), and White House PCAST processes.
- Lead author and coordinator on the forthcoming UNESCO/UN AI Science Report, and co-author of the MIT AI Safety Repository.
- Evaluator, rapporteur, and quality controller across EU Horizon, StandICT, and EIT Health, Deeptech, and Creativity programmes; member of the EDPB expert pool on embodied AI privacy.
- Former curator of the Global AI Summit, and speaker at the AI Action Summit (Paris/Sorbonne), Sciences Po, Technical University of Munich, AI Safety Korea, and the Delphi Economic Forum; jury and peer reviewer for the MIT Global Competition, NeurIPS, and IEEE journals.
Biography
The EU AI Act, the OECD framework, UNESCO’s ethics recommendation, and the G7 Hiroshima Process now govern the same operating decisions, and they do not line up neatly. The people who work inside several of these processes at once, rather than reading them from outside, are few.
Welker is one of them. He has contributed to the EU AI Code of Practice working groups and to DG CNECT’s critical infrastructure work. He sits on the EDPB expert pool examining embodied AI privacy, and is lead author and coordinator on the forthcoming UNESCO/UN AI Science Report. His input has also fed NIST and White House PCAST processes.
His distinctive angle is a lens few governance specialists share. His work centres on how AI systems treat designated groups: people with disabilities, neurodivergent users, and others where an algorithm can quietly grant or deny access. That question is now moving into robotics, assistive systems, and AI that acts in public space, where the frameworks are thinnest. The disability-centred AI MOOC catalogued in the OECD.AI tools repository is one concrete marker of it.
For senior leaders, the value is operational. He can set out what an EU AI Act audit will actually examine, and what a defensible accessibility claim looks like in a deployed system. He can also show how the OECD, UN, and EU layers connect to a decision a general counsel or risk committee faces next quarter.
Key speaking topics
- AI governance and the EU AI Act
- Algorithmic accountability and AI audit
- Disability-centred and accessibility AI policy
- Neurodiversity and inclusive AI design
- Sovereign AI and critical digital infrastructure
- AI safety and responsible deployment
- Public sector AI and citizen-facing systems
Ideal for
- General counsel, chief compliance officers and CTOs preparing for EU AI Act enforcement
- Heads of AI, responsible AI leads and digital ethics committees
- Public sector and intergovernmental bodies designing AI policy
- Boards reviewing AI risk exposure across accessibility, bias and public-trust dimensions
Audience outcomes
- A clearer map of how the EU AI Act, OECD framework and UN/UNESCO instruments interlock for their organisation
- Specific questions to put to vendors, internal teams and audit committees on algorithmic accountability
- A working definition of what disability-centred and accessibility AI obligations look like operationally
- Sharper instinct for where reputational and regulatory risk concentrates in current AI deployments
- A view of where European AI policy is heading next, drawn from inside the working groups shaping it
Talks
A working session on what human-centred AI means once disability, neurodiversity and accessibility move from principle to compliance obligation.
Key takeaways:
- How algorithmic decisions affect designated groups and where audit attention should focus
- Practical anchors from the OECD.AI tools catalogue and UNESCO ethics work
- What inclusive AI design looks like in product, procurement and public service contexts
A senior-level walk through the overlapping AI governance frameworks now shaping organisational decisions.
Key takeaways:
- The current state of the EU AI Act and adjacent codes of practice
- How OECD, G7 Hiroshima Process and UN IGF outputs translate into board-level questions
- What a defensible AI governance posture looks like in practice, beyond policy statements