Download ORBICAP AI Policy, version 1.1 (PDF)
ORBICAP AI Policy
Version 1.1, August 2025
ORBICAP's AI Policy establishes principles and guidelines for the ethical, effective, and responsible use of Artificial Intelligence (AI) tools and technologies across its operations. This policy aligns with international standards and best practices, ensuring ORBICAP’s commitment to transparency, accountability, and privacy while leveraging AI's potential to enhance services in leadership development, workflow optimisation, quality assurance, and data-driven decision-making.
1. Purpose and Scope
This policy outlines ORBICAP's approach to adopting and endorsing AI tools in its operations. It applies to all employees, contractors, and collaborators utilising AI tools and systems. The endorsed tools and their recognised users are detailed in the attached appendix, 'Comprehensive List of Endorsed AI Tools.'
This policy also supports ORBICAP’s training programmes, including regional courses on statistical quality and academic lectures on official statistics, by ensuring that AI tools are used transparently and responsibly to enhance data integration, documentation, and analysis.
2. Guiding Principles
The following principles guide ORBICAP's AI use and endorsements:
o Ethical Use: AI tools must respect ethical standards, human dignity, and organisational values.
o Transparency: Where applicable, clients, partners, and stakeholders will be informed of AI usage, and disclosures will be provided in deliverables.
o Privacy and Compliance: All data processed using AI tools must comply with local and international regulations, including GDPR, and user data must be anonymised and secure.
o Accountability: Humans will review AI outputs to ensure accuracy, cultural sensitivity, and alignment with project objectives.
o Alignment with International Standards: ORBICAP endorses AI tools recognised by international organisations such as PARIS21, the United Nations, the World Bank, the IMF, and Eurostat. The appendix provides specific tools and their uses.
3. Endorsed Tools and Usage
ORBICAP endorses a range of AI tools because they align with best practices and can be used in leadership, workflow, and quality-focused projects. The appendix details these tools and their recognised users, including their applications in data visualisation, transaction analysis, and mobile data collection.
4. Implementation Guidelines
The use of AI tools at ORBICAP will adhere to the following implementation guidelines:
o Training and Capacity Building: All staff will be trained to use endorsed AI tools responsibly.
o Tool Selection: AI tools will be selected based on project requirements, client needs and alignment with applicable regulations such as GDPR and the Statistics Act. Tools must support statistical quality, protect privacy, and demonstrate practical value in data-driven decision-making. Only tools listed in the appendix or approved by ORBICAP leadership will be used.
o Client Communication: When AI tools significantly contribute to project deliverables, clients will be informed. This disclosure will be made through reports or presentations with a dedicated section in quality documentation templates (e.g., ‘AI Tools Used: Power BI for visualisation, validated by ORBICAP experts’). Example statements include:
“Data visualisation created using Power BI.”
“Missing values imputed using DataRobot, with expert validation.”
“Survey data collected with Kobo Toolbox integrated for SDG monitoring.”
o Monitoring and Review: AI use will be reviewed annually during Q1 to ensure continued compliance with ethical standards, the EU AI Act, GDPR, and international statistical frameworks.
5. Risk Management
ORBICAP will actively manage risks associated with AI usage, including:
o Identifying and mitigating potential biases in AI outputs.
o Ensuring that all data processed using AI tools is anonymised and secure.
o Maintaining human oversight to validate and refine AI-generated outputs.
6. Legal Compliance and the EU AI Act
ORBICAP does not develop or distribute AI systems. We act as a professional user of selected, externally developed AI tools in support of leadership, quality, and data-focused consulting. Following the EU Artificial Intelligence Act (AI Act), ORBICAP classifies its use of AI as low-risk and ensures complete transparency, human oversight, and privacy protection in all relevant applications.
When advising public institutions or statistical producers on the use of AI in data workflows or capacity building, ORBICAP promotes alignment with the EU AI Act, the UN Fundamental Principles of Official Statistics, and the European Statistics Code of Practice. This includes highlighting when specific AI applications—such as classification, imputation, or editing—may fall under high-risk categories according to the Act, and ensuring these are managed accordingly through robust documentation, ethical review, and stakeholder communication.
7. Appendix
The attached appendix, ' Comprehensive List of Endorsed AI Tools, ' provides a detailed list of endorsed AI tools, their recognised users, and specific applications. This appendix serves as a reference for understanding the scope and utility of these tools.
Appendix: Comprehensive List of Endorsed AI Tools
1. General AI Tools for Workflow Optimisation and Quality Assurance
o ChatGPT (OpenAI): Used by the World Bank for drafting reports, automating communication, and summarising policy documents.
o Jasper: Recognised by PARIS21 for generating high-quality content and training materials for statistical capacity-building projects.
o Notion AI: Eurostat adopted it for workflow optimisation and internal task automation.
o UiPath: Deployed by the IMF to automate repetitive data processing tasks in economic research workflows.
o Power BI: Widely used by the United Nations for real-time data visualisation and SDG reporting.
o Tableau: Leveraged by the OECD for interactive dashboards and predictive data analysis in policymaking contexts.
2. Endorsed Tools for Mobile Phone Data Collection and Analysis
o FlowKit (by Flowminder): Used by the World Bank to analyse population mobility and health trends in low-income countries.
o Mobility Insights (Google): Adopted by the United Nations to track mobility patterns during global emergencies like COVID-19.
o Orange Data for Development (D4D): Supported by the United Nations Global Pulse initiative for analysing development challenges.
o Positium Data Solutions Collaborates with Eurostat to provide insights into migration and tourism statistics using mobile network data.
o H3 Geo Data Framework: Recognised by the World Bank for spatial analysis in urban development projects.
3. Endorsed Tools for Transaction Data Collection and Analysis
o World Bank Microdata Library: The World Bank maintains a platform for accessing datasets and supporting economic policy development.
o Mastercard Data Insights: Used by the IMF for analysing consumer spending and economic trends in emerging markets.
o Visa Economic Empowerment Tools: Recognised by the OECD for financial inclusion studies and economic empowerment initiatives.
o SafeGraph: Adopted by the World Bank to understand market trends and consumer behaviour using transaction data.
o DataRobot: Leveraged by the IMF for predictive modelling in financial and macroeconomic analyses.
4. Tools Recognised by International Organisations
PARIS21:
o ADAPT: Developed and used by PARIS21 to align statistical systems with development priorities such as the SDGs.
o Kobo Toolbox: The United Nations uses this toolbox to collect field survey data and supplementary datasets.
United Nations (Big Data for Official Statistics):
o AI-driven platforms for analysing satellite, mobile, and financial data in collaboration with the World Bank.
Kobo Toolbox: The United Nations uses this toolbox to collect field survey data and supplementary datasets.
World Bank:
o Pulse Data Hub: The World Bank uses real-time mobile and transactional data to monitor economic shocks.
Eurostat (ESS):
o Guidelines for mobile network data use in experimental statistics, supported by European national statistical offices.
o Pilot projects integrating AI into transactional and mobile phone data analysis.
5. GIS-Related AI Tools
o ArcGIS: Used by the World Bank and United Nations for geographic data visualisation and spatial analysis.
o QGIS (Quantum GIS): Recognised by Eurostat for spatial data mapping and analysis in national statistical offices.
o Google Earth Engine: Leveraged by the United Nations for geospatial data processing in environmental and urban studies.
o GeoDa: Adopted by the OECD for spatial data analysis in regional policy development.
6. Additional Statistical and Analytical Tools
o RapidMiner: Used by the IMF for statistical modelling and advanced analytics in macroeconomic studies.
o TensorFlow/Keras: Applied by the OECD for developing machine learning models in predictive analytics projects.
o PySAL (Python Spatial Analysis Library): Eurostat adopted this library for spatial analysis and modelling in urban and regional planning.
7. ORBICAP Implementation Notes and Legal Position
Alignment with Best Practices: ORBICAP endorses these tools for their alignment with international standards.
Focus Areas: Tools are categorised for workflow optimisation, quality assurance, analysis, and specialised data collection.
Client Projects: The selection of tools will depend on the project requirements, and clients will receive recommendations based on their needs.
Legal Scope: ORBICAP uses AI tools only following applicable laws and regulations. As an AI user (not a developer or deployer), ORBICAP falls under the obligations for low-risk applications in the EU AI Act and complies with its transparency and oversight provisions.
8. Glossary of Key Terms
To support accessibility for non-technical audiences, the following terms are used in this policy:
Imputation: Filling in missing data using statistical methods, e.g., estimating missing values in regional healthcare data.
Classification: Assigning categories to data based on patterns, e.g., grouping survey responses.
Transaction Analysis: Analysing financial or consumer data, e.g., spending trends for regional economic planning.
Predictive Modelling: Forecasting trends using data patterns, e.g., predicting healthcare demand.
Statistical Quality Documentation: Describing strengths and limitations of statistics using criteria like accuracy or comparability, per the Statistics Act.
SDG: Sustainable Development Goal, a UN-defined development target (e.g., health, education)
This page was updated in August 2025.