Topics
At this year’s Közgazdász-vándorgyűlés (the Hungarian Economic Association’s annual conference), held in Eger, two sessions explored Europe’s AI challenge. One examined whether Europe can build and deploy the technology it depends on. The other asked how organisations can keep increasingly autonomous systems accountable and under human control.
Representing E-Group, Antal Kuthy addressed Europe’s strategic position, while Ferenc Pongrácz demonstrated COIOS and took part in a separate discussion on AI accountability.
Superpower ambitions and the mindset of a middle power
In the session “Who Will Win the AI Revolution in Hungary?”, Kuthy, co-founder of E-Group, argued that Europe needs to act like a superpower, yet still thinks like a middle power. He also serves on the IPCEI-CIS Facilitation Group and is a workstream leader. IPCEI-CIS is a European initiative that brings companies and governments together to develop next-generation cloud and edge computing technologies and strengthen Europe’s digital sovereignty.
The scale of Europe’s cloud dependence helps explain that concern. In estimates published in July 2025, Synergy Research Group put the combined share of Amazon, Microsoft and Google at around 70% of Europe’s cloud infrastructure services market.
Kuthy captured his argument in a memorable line, translated from Hungarian: “If you cannot forge your own sword, you have no sword.” Europe’s research base and technical talent, he argued, must translate into competitive technology that organisations can deploy on infrastructure under their own control.
Watch Antal Kuthy’s remarks from 30:12
Sovereignty in practice
Pongrácz presented a live demonstration of COIOS, E-Group’s sovereign AI platform.
The illustrative scenario followed a finance minister’s working day: preparing a press review, identifying a sudden oil-price risk, modelling its budgetary impact, assessing potential social consequences and drafting a government proposal. In the demonstrated configuration, data and AI processing remained within the locally hosted environment on Hungarian infrastructure, without using a hyperscaler cloud.
The demonstration illustrated how AI can connect information gathering, analysis and document preparation within a single workflow. For institutions handling sensitive information, the ability to run that workflow on infrastructure under their own control is central to the case for sovereign AI.
Watch Ferenc Pongrácz’s COIOS demonstration from 38:08
Robots on a leash
In “Robots on a Leash: The Role of SLAs”, a prerecorded roundtable included in the conference programme, Pongrácz joined a discussion moderated by Marianna Ritter about how organisations can govern increasingly autonomous AI systems.
A service-level agreement can set expectations for availability and performance. AI agents introduce additional questions about permitted actions, supervision and responsibility when something goes wrong.
For organisations deploying AI, the discussion points to a practical requirement: define what an agent is authorised to do, when a person must approve an action, and how the system’s activity can be traced and reviewed. Contracts, technical safeguards and operational governance need to work together.
Pongrácz brought Europe’s technology dependence into this discussion, linking the governance debate to the infrastructure and providers on which AI runs.
Watch the “Robots on a Leash” roundtable
The two conference themes meet here. Control over hosting and data provides a foundation for sovereign AI. COIOS is designed to support local deployment, access controls, human oversight and an auditable record of system activity. These capabilities can help organisations apply their own governance policies.
Building at European scale
The wider challenge is to make these capabilities available at European scale. Initiatives such as IPCEI-CIS create a framework for companies to develop interoperable cloud and edge technologies that European organisations can adopt.
Kuthy’s call was for such efforts to be replicated far more widely across Hungary and Europe. For Europe, the task is to build AI it can use at scale and ensure that the people and institutions using it remain in control.
Sources:
https://news.bpdata.com/hu/article/szerzodesek-es-felelosseg-az-onalloan-mukodo-ai-rendszerek-ec1e839c
https://kozgazdasz-vandorgyules.blog.hu/