World Bank report on AI in Romania’s pre-university education cites LIVRESQ
The World Bank Education Global Practice published “Artificial Intelligence in Romania’s Pre-University Education”, mapping the national AI and EdTech landscape. LIVRESQ is documented as one of the country’s authoring platforms, with an Ascendia contributor acknowledged among the report’s consulted stakeholders.
Edited by Gabriel Lazar — Research & Development Director · EC DEC expert-group member · gabriel.lazar@ascendia.eu
The World Bank published “Artificial Intelligence in Romania’s Pre-University Education”, Brief No. 8 in the Digital Innovations in Education series, produced by the Education Global Practice. The report maps how artificial intelligence is currently used across Romania’s pre-university system, assesses national readiness for AI adoption, and sets out six recommendations for policymakers.
The study was prepared by Lorena Chiran under the guidance of Cristobal Cobo and Alina Sava, Senior Education Specialists. Its methodology rests on semi-structured interviews conducted between January and February 2026 with 21 stakeholders drawn from government, the private sector, non-governmental organisations, Romania’s AI Hub and a rural school, complemented by platform documentation and secondary data on digital infrastructure, educational performance, digital skills and EdTech adoption.
Ascendia is named in the report’s acknowledgements: Gabriel Lazar, Project Manager at Ascendia, is listed among the stakeholders whose collaboration, insights and feedback informed the work. The report also carries a quoted contribution from Ascendia’s Research and Development function on the conditions for effective classroom use of AI: “AI is a working tool, and if used incorrectly, it generates negative results. That is why it must be used effectively, and for that, teachers must learn to use it and to teach their students to use it too.”
LIVRESQ appears in the report’s inventory of AI tools supporting teachers. The World Bank describes it as an authoring platform enabling teachers to create interactive, curriculum-aligned digital lessons, including an AI Enhancer built on Azure OpenAI that generates, rewrites or adapts lesson texts, dialogues and content by age, tone, lesson duration or keywords. The stated function is to accelerate lesson planning, support the creation of high-quality digital teaching materials, and personalise instructional content.
The report describes LIVRESQ as hosting the largest online collection of Romanian educational materials, officially recommended by the Romanian Ministry of Education and serving as the primary tool for the national EduLib project and the CRED initiative. Its annex records the platform at more than 130,000 teachers across 7,100 institutions, over 120,000 registered content creators and more than 89,000 library publications, including 20,596 free public lessons.
Those annex counts are a point-in-time snapshot rather than current scale. They reflect the platform documentation available during the report’s research window, which closed in February 2026. Official Ascendia figures record 194,756 materials published in the LIVRESQ library, more than 160,000 registered creators and more than 7,200 institutions — more than double the published library volume the report documents, reached in the months since its research closed.
In its narrative sections, the report positions LIVRESQ within a Romanian EdTech market it characterises as sophisticated and increasingly scalable, alongside Adservio, EDUS.ro, BRIO, Digitaliada, Clasa Viitorului and LearnMate. The World Bank credits this group with strong curriculum alignment, integration with national quality-assurance frameworks, and a deep understanding of Romanian school realities — evidence, in the report’s framing, of national capacity to adapt AI to Romanian-language accuracy and context-specific pedagogical constraints.
The report is not an endorsement of any single provider, and Ascendia does not present it as one. Its central finding is a systemic one: ecosystem readiness in Romania is strong but fragmented. Capable platforms exist and are used at scale, but interoperability standards, public–private procurement pathways and shared evidence standards are not yet established, so impact remains localised rather than systemic. The report states this directly in relation to authoring: LIVRESQ supports AI-assisted lesson authoring at national scale, yet the surrounding ecosystem remains loosely connected to curriculum and lacks shared evidence standards.
That diagnosis is the part of the report most relevant to Ascendia’s institutional work. The recommendations call for national-level guidance on AI integration aligned with the National AI Strategy and the EU AI Act, a permanent AI-in-Education Taskforce that includes Romanian EdTech providers and maintains a vetted tools list, professional development reoriented from training delivery toward classroom application, and public–private research and innovation pipelines with “test-before-scale” mechanisms governed through the AI Hub and European Digital Innovation Hubs.
Those are the same conditions Ascendia works under in national and European delivery: curriculum-aligned authoring, accessibility and interoperability as procurement criteria, measurable classroom application rather than completion counts, and a documented evidence trail. The company’s participation in the European Commission Digital Education Content expert group addresses the quality-criteria half of the problem; the World Bank report describes the procurement and evidence-standard half that remains open in Romania.
Ascendia links to the World Bank publication record rather than hosting a local copy, so readers can consult the current official document, its citation and its terms of use directly.
Evidence and contacts
