Experts examined how generative and agentic AI systems can be aligned with human values and public interest.

A panel on AI Accountability in the Public Sector wrapped up the first day of The Responsible AI Forum on May 19th, 2026. Moderated by Public Interest AI track chairs Nicole Manger (TUM Think Tank) and Dalia Yousif Ali (Technical University of Munich), panel participants included Dr. Auxane Boch, Dr. Caitlin Corrigan and Prof. Laura Crompton. Together, they dug deeper than conventional discussions on AI ethics, for an honest and in-depth discussion on AI and the realities of implementation, governance trade-offs, human rights, inclusion and democratic legitimacy.

The Role of Public Institutions

The panel kicked off by first looking at the subject through a broader lens. Panelists weighed in on how public institutions can leverage the field, a misalignment in public interest and market incentives, what counts as public interest as well as structural challenges. They were then given questions based on their individual expertise.

When asked where we are falling short and what has the greatest potential to drive meaningful societal change, Dr. Corrigan, Managing Director of TUM TransforM, stressed that to drive real, meaningful change, it requires the active implementation of agreements and backing them with financial support.

Prof. Crompton, AI Ethics Professor at the Technische Hochschule Ingolstadt, cited the public sector’s tendency to work in silos. A major problem with this is a lack of transparency. There is an urgent need to work horizontally.

Defining Public Interest

A core moderation question of the next discussion block was “Who gets to define ‘public interest’ in AI governance—and what makes that definition legitimate?”. A focus was placed on ‘public interest’ as a term frequently invoked to justify AI deployment in healthcare, surveillance and public services, yet it is vague itself in terms of definition. Additionally, panelists spoke on who defines ‘public interest’.

Dr. Boch, Associate Research Director of The Institute for Ethics in Artificial Intelligence, shared her insights: “I’ve seen fear and hype inform policy, but not data”. Dr. Boch continued, underscoring the importance of scientists playing a larger role and their involvement throughout the entire process.

Noting AI’s growing influence on decision-making, Prof. Crompton reminded the public sector has a different kind of responsibility to consider. The public sector is, at its core, created to serve the people. With an increasing influence from AI, there is a mismatch in this regard.

Operationalising Accountability

Next, panelists debated which governance mechanisms make accountability enforceable and how public institutions can operationalise human rights and accountability considerations across the whole AI life cycle.

Prof. Crompton gave an insightful explanation from a German perspective when asked about institutions and accountability. In German, responsibility can be translated as “Verantwortung” (responsibility) or “Zuständigkeit” (role responsibility). Role responsibility is less significant than responsibility and therefore has implications on how much agency is granted. Less significance equals less decision-making power, which in turn affords a lower amount of accountability. Institutions can potentially shirk being held accountable through this nuance.

Dr. Boch then stressed the fundamental importance of understandability. Education should be an accessible prerequisite for all—this is at the core of what it means to be democratic. She further stressed the value of building globally.

When asked how governments can ensure meaningful public participation from the most affected communities, Dr. Corrigan encouraged a concerted effort to get a global baseline. She further highlighted the process as key, with a binding framework not necessarily being the end goal.

Closing Remarks

Lastly, panelists were asked what one concrete step public institutions could take in the next year to advance AI aligned with public interest. The answers were a clearer definition of public interest, funding and change in the procurement process. The floor was then opened for questions from the audience.

Watch a video of the panel below.