Since 2019/20 Croatia's Ministry has funded a national Turnitin licence via SRCE for all public universities; checks grew from 74,711 to 265,106 papers by 2023/24. A 2024/25 AI-text detector covers only English/Spanish, and SRCE warns it can't definitively prove AI use.
~1.9 million kuna
Initial Ministry funding (AY2019/2020)
68 institutions
Institutions actually using the software (launch year) (AY2019/2020)
70 institutions
Institutions using licensed software (steady years) (AY2020/21–AY2022/23)
Annual support enquiries handled by SRCE (ongoing)
Details
Maturity
Established
Promoter
SRCE (University of Zagreb University Computing Centre) / Croatian Ministry of Science, Education and Youth
Period
2019–2025
Keywords
higher education, academic integrity, AI-text detection, plagiarism detection, national infrastructure
Context
Since academic year 2019/2020, Croatia's Ministry of Science and Education (now Ministry of Science, Education and Youth) has centrally financed unlimited-use plagiarism-detection software licences for every public higher-education institution, coordinated technically and pedagogically by SRCE, the University of Zagreb's University Computing Centre.
Objectives
Provide a national, centrally funded plagiarism- and (from 2024/25) AI-text-detection infrastructure so that every public higher-education institution can check student work, with standing training and student self-check access.
Activities
At launch the Ministry reported roughly 1.9 million kuna in funding covering 72 institutions; SRCE's own usage data record 68 institutions actually using the software that first year, 52.94% choosing Turnitin over the since-discontinued PlagScan. In AY2024/2025, Turnitin's AI-generated-text detection module was added to the national licence, but only for papers written in English and Spanish, not Croatian.
Results
Usage scaled steadily: 70 institutions used the licensed software in each academic year from 2020/21 to 2022/23, and papers checked rose from 74,711 in AY2019/2020 to 265,106 in AY2023/2024 — a 64.65% increase between AY2020/2021 and AY2023/2024 alone. By AY2023/2024, 62 of 70 public higher-education institutions (89%) had enabled students to self-check their own papers before submission, and SRCE fields roughly 300 support enquiries per year alongside standing 'Authorship, Plagiarism, and Citation' training courses.
Conclusions
SRCE researchers are explicit about the AI-text detector's limits: it 'cannot definitively assess whether a paper or a portion of it was written using AI; instead, they primarily indicate the possibility of AI involvement,' and results require human interpretation in disciplinary context — a candid limitation rather than an oversold accuracy claim.
Implementation
Indicative cost
Medium (€50k–€500k) — Roughly 1.9 million kuna in initial Ministry funding (AY2019/2020) for a national unlimited-use licence, scaled to 70 institutions and rising usage volumes in subsequent years — conservative medium estimate pending curator review.
Time to results
Long (> 3 years) — Running continuously since AY2019/2020 through at least AY2024/2025 (six academic years), with the AI-text module added in AY2024/2025 — a long-running national programme.
Staffing & skills
SRCE (University of Zagreb University Computing Centre), Croatian Ministry of Science, Education and Youth
Conditions for success
Centralised national procurement/licensing model covering every public higher-education institution
Standing 'Authorship, Plagiarism, and Citation' training for staff and students
Student self-check access enabled at 89% of institutions by AY2023/2024
SRCE fields roughly 300 support enquiries per year
Common failure modes
AI-generated-text detector covers only English and Spanish, not Croatian
SRCE explicitly cautions that AI detectors cannot definitively assess AI involvement and require human interpretation
Where it fits
Governance type
national ministry-funded programme
Scale
national (all public higher-education institutions)
Commonly funded by
National / regional programmes
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
Facing evidence that AI-text detectors misclassify non-native English writing, South Africa's UCT, Stellenbosch, UFS and Wits disabled Turnitin's AI detector …
Grenoble-based plagiarism and AI-content detection platform used by 1,100+ institutions in 50+ countries; LMS-integrated and GDPR-compliant, with 4M+ students reached …
Vanderbilt disabled Turnitin's AI-writing detector campus-wide in Aug 2023, calculating its claimed 1% false-positive rate could wrongly flag ~750 of …