Technology 4 min read By Callum Montgomery
Romania to host first national AI infrastructure, targeting full operation in 2027
ICI Bucharest is leading the development of Romania's first national AI Factory, a project linked to a European network, with completion targeted for 2027 and officials stressing data sovereignty and human control over financial decisions.
Romania is moving towards its first national artificial intelligence infrastructure, with the AI Factory project led by the National Institute for Research and Development in Informatics (ICI) Bucharest as host entity. The initiative is designed to be fully operational by 2027, although the notes available do not confirm that this timeline will hold.
The project is part of a European consortium agreement that has already been signed, and it includes the Politehnica University of Bucharest among its partners. Its stated aims are to enable the processing of confidential data while preserving data sovereignty, so that sensitive information remains under national control rather than being handled entirely by foreign platforms. The infrastructure is also meant to support research, specialist training and practical applications, and to connect Romania to the wider European AI infrastructure network, allowing domestic institutions to take part in shared capacity across the bloc.
Cybersecurity was flagged during the discussion as a critical concern, given the sensitivity of the data that such systems may handle. The stakes are particularly high when confidential datasets are involved, since a breach could expose information belonging to institutions, companies or citizens. Participants also floated the idea of an emergency shutdown mechanism for autonomous AI agents, a safeguard that would allow human operators to halt systems that behave unexpectedly or outside their intended parameters.
The debate was framed largely around the financial sector, where officials said that final decisions in AI-assisted finance must remain in human hands. The dean of the Politehnica University of Bucharest argued that artificial intelligence is not yet ready to make independent decisions, a view that shaped much of the discussion on governance and oversight. The position reflects a broader caution among academics and regulators that automated systems can be useful for analysis while still requiring accountable human judgement at the point of decision.
The remarks were made at the Digital Finance forum in Bucharest, organised by the Financial Supervisory Authority (ASF) together with the Institute of Financial Studies. The financial regulator already uses AI for data processing, and the situation is mixed: some of its tools are operational, while others are still in testing. Among the functions already in use are the detection of unusual market behaviour and the analysis of company disclosures, tasks that involve scanning large volumes of information far faster than manual review would allow.
The article from which these notes are drawn does not state the exact funding amount for the AI Factory, so the financial scale of the project and the sources of its budget remain unclear. Equally, no specific timetable for the intermediate stages of development is given beyond the 2027 target. The combination of a national infrastructure, a European network connection and a regulator already experimenting with the technology places Romania's AI programme at the centre of ongoing questions about how public institutions should balance innovation with control, security and accountability.
Source: economedia.ro
4
