Morocco's Directorate General of Taxes (DGI) operates a bilingual AI chatbot (DGIBot, launched January 2022) on its web portal (tax.gov.ma) and Daribati mobile app — handling 11,169 conversations in 2024 — alongside a ML-based Système d'Analyse des Risques (SAR) that cross-refere
11,169 conversations
DGIBot conversations handled (2024)
9 %
Share of total customer requests handled by chatbot (2024)
17.77 billion MAD (~$1.76bn)
Additional revenue from tax controls (2024)
26 %
Year-on-year increase in tax control revenue (2024)
7,674 cases
On-site audit cases (2024)
32 %
Growth in on-site audit cases (2024 YoY)
242.48 billion MAD
Gross tax receipts (2024)
16.1 %
Growth in gross tax receipts (2024 YoY)
Details
Maturity
Established
Promoter
Direction Générale des Impôts (DGI), Ministry of Economy and Finance, Morocco
Morocco's Direction Générale des Impôts (DGI) runs a dual AI system for tax administration: DGIBot, a bilingual (Arabic and French) chatbot available 24/7 on the tax.gov.ma portal and Daribati mobile app since January 2022, and the Système d'Analyse des Risques (SAR), a machine-learning system that cross-references DGI's fiscal database with customs (ADII), social security (CNSS, CNOPS) and foreign-exchange records to flag anomalies and non-compliance. The initiative is governed under Law 55-19 on administrative simplification.
Objectives
The system aims to improve citizen access to tax-administration services through the chatbot channel, and to detect tax fraud and non-compliance more effectively — and thereby raise tax revenue — through automated risk analysis.
Activities
DGIBot handles routine citizen queries around the clock in Arabic and French, while SAR automatically cross-references multiple government databases to flag anomalies for follow-up by tax inspectors, who then conduct targeted audits.
Results
In 2024, DGIBot handled 11,169 conversations, about 9% of DGI's 113,119 total customer requests. Tax controls generated 17.77 billion MAD (about $1.76 billion) in additional revenue, a 26% year-on-year increase, on-site audit cases grew 32% to 7,674 cases, and gross tax receipts reached 242.48 billion MAD, up 16.1%.
Conclusions
These are administrative monitoring figures rather than results of a controlled evaluation, and DGI's own reporting does not isolate how much of the revenue and audit growth is attributable to the AI risk-analysis system specifically versus increased inspector capacity and other factors.
Implementation
Indicative cost
High (€500k–€5M) — No public budget figures were found; building and integrating a chatbot plus a cross-database ML risk-analysis system spanning tax, customs, social security and forex records nationally implies substantial systems-integration cost, estimated as high in the absence of disclosed figures.
Time to results
Long (> 3 years) — The AI initiative dates to around 2019, with DGIBot launched in January 2022 and SAR/reporting continuing through at least 2024/2025 — a multi-year, long implementation and operating horizon.
Staffing & skills
Direction Générale des Impôts (DGI) (system ownership and operation), Customs administration (ADII), social security funds (CNSS, CNOPS), and the foreign exchange office (data-sharing partners for SAR), Tax inspectors (follow-up audits triggered by SAR-flagged anomalies)
Conditions for success
Legal basis for administrative data-sharing and simplification under Law 55-19
Interoperability across DGI's fiscal database, customs, social security and forex records for SAR to cross-reference
Bilingual (Arabic/French) chatbot interface integrated into existing tax.gov.ma portal and Daribati mobile app
Sustained inspector capacity to act on SAR-flagged non-compliance cases
Common failure modes
DGI's own reporting does not isolate the AI system's specific contribution to revenue/audit growth from increased inspector capacity or other factors
Chatbot handles only a modest share (~9%) of total customer requests, suggesting most citizen interactions still occur through other channels
No independent or third-party evaluation of the risk-analysis system's accuracy or fairness was found in accessible sources
Where it fits
Governance type
national tax administration
Scale
national, whole tax base
Income level
lower-middle-income (Morocco)
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.
Sierra Leone's National Medical Supplies Agency and University of Pennsylvania researchers built a decision-aware ML tool that raised essential-medicine consumption …