The pilot project seeks to improve the data and knowledge related to incidents of illegal fishing across the world. The objective was to use innovative techniques to identify and collect new data. Generative AI’s deep research capabilities were used to identify and expand the range of relevant data sources. AI-assisted tools were used to verify and adapt to the varying structures of web pages, ensuring the web scraping process remains robust despite differences in page layouts and metadata formats across sites.
Activity Type Research/Reports/AssessmentsThe UNODC Illicit Crop Monitoring Programme (ICMP) uses GIS and geospatial analysis, satellite imagery and field surveys to monitor the extent and evolution of illicit crop cultivation and production. The project seeks to research and eventually apply (semi-) automated methods such as deep learning and big data analysis for improving area estimates for illicit crop cultivation. Moreover, research is conducted on spectral based yield information, and the early detection of illegal landing runways applying AI techniques.
Activity Type Research/Reports/AssessmentsAI Tools/Solutions
Business-as-usual makes people with disabilities pay for exclusion, treating it as a technology gap. We call it what it is: a design failure.
We didn’t ask for a seat; we rebuilt the table. Co-created by UN Volunteers with lived experience, our zero-budget engine delivers 52 features and WCAG 2.2 AAA compliance through a single interface—allowing users to adapt content to their abilities.
This shifts accessibility from an individual burden to a core capability, pushing the UN from theoretical commitments to practical, universal inclusion.
The problem:
Many persons with disabilities still struggle to access online information, including on UN platforms. This contradicts UN international commitments and accessibility charters. While UN information exists, access remains unequal. Current accessibility solutions are fragmented (one tool per disability) and often costly or beyond users’ budgets. Responsibility is shifted from institutions to individuals, forcing people to pay to exercise a basic right. This model excludes people with multiple disabilities and undermines participation and trust.
The solution:
UN AI Accessibility is a paradigm-shifting multi-disability system designed by persons with disabilities, for real human beings. Instead of forcing users to adapt, the system adapts to them.
That is the vision behind UN AI Accessibility—a multi disability system, designed by persons with disabilities, for real human beings. Instead of forcing users to adapt, the system adapts to them.
It is a single, lightweight, zero budget solution that supports visual, motor, cognitive, auditory, and language disabilities—all within the same interface and on the same UN content. Users can read, listen, navigate, understand, simplify, or use sign language according to their abilities.
The system is already functional and ready to deploy pending IT governance and root-level authorization, turning accessibility from a promise into practical, human centred digital access.
The innovation :
The UN AI Accessibility initiative is an outstanding systemic innovation shifting digital accessibility from a private burden to a core UN capability. Designed by UN Volunteers with disabilities, it turns 'nothing about us without us' into real practice.
Concretely, it delivers a unified engine of 52 features for 8+ disability profiles, achieving the highest standard (WCAG 2.2 AAA) with a zero-dollar budget. Instantly deployable via one HTML file, it proves inclusion is a design choice, not a cost.
Only the UN, with its unique mandate and legitimacy, can scale such a paradigm shift. By absorbing this responsibility, the UN operationalizes its values—turning rights into practical access—and makes digital inclusion credible, scalable, and universal. It is the promise of a UN truly accessible for all.
The potential impact :
To date, the initiative has operated as a fully functional, highly advanced prototype, achieving significant efficiency gains, cost-saving, and technical validation:
1. Unprecedented Resource & Time Savings:
$0 Budget Foundation: Developed entirely in-house by UN Volunteers, the project bypassed the massive procurement costs typically required for accessibility audits or fragmented commercial software.
Process Simplification: We reduced the overwhelming complexity of digital compliance down to one single HTML file. It requires zero rewriting of existing UN website source codes, significant reduction in integration effort.
2. Technical Achievement & Stakeholder Validation:
System Capability: We successfully engineered a unified engine with 52 distinct features, catering to 8+ disability profiles, strictly adhering to the WCAG 2.2 AAA standard.
Internal Buy-in: The solution has been demonstrated to internal UN teams, generating a 'Wow effect' and positive feedback for its 'one interface, multiple pathways' design.
3. The Catalyst for Scale:
Institutional Deployment: The engine currently runs flawlessly in a localized environment. To unlock our ultimate metric—reaching all UN web users—we need IT governance support to authorize placing the file in the root directories of UN websites.
Unlocking Premium AI Power: While the core engine was built at zero cost, realizing its ultimate potential requires enterprise access to paid AI APIs (e.g., Google Gemini). UN 2.0 support would allow us to integrate these premium models, unlocking next-generation, dynamic accessibility features and making the tool unprecedentedly performant.
Impact to date is qualitative and validated; the scale impact depends on deployment support.
A four-day in-person training course equipping law enforcement officers, prosecutors, and judges with practical AI skills for criminal investigations. Participants learn to use AI tools to identify suspects, analyze digital evidence, and build structured profiles using facial recognition, language models, and link analysis techniques.
Activity Type Trainings/Workshops