By Alexandre Essome and Taurai Chiraerae

The 2025 G20 Summit hosted by South Africa from 22 – 23 November was the first ever held on African soil and placed illicit financial flows (IFFs) at the centre of the global economic agenda. Under the theme “Solidarity, Equality, Sustainability,” the Johannesburg Leaders’ Declaration explicitly recognised that IFFs “constitute one of the most significant barriers to Africa’s sustainable development” (G20, 2025a, para. 27). Annual losses are conservatively estimated at USD 88.6 billion (UNCTAD, 2020) and may now exceed USD 120 billion when cryptocurrency and intangible asset transfers are included (Chainalysis, 2025).

The Summit produced the Johannesburg Leaders’ Declaration, a Ministerial Call to Action containing ten high-level principles for combating IFFs, a renewed G20 Africa Engagement Framework (2025–2030), and the Ubuntu Legacy Initiative all of which explicitly encourage the responsible use of artificial intelligence and digital technologies (G20, 2025a, 2025b). While these commitments are historic, they remain voluntary. Turning political will into measurable impact now depends on rapid, African-led technological deployment. Artificial intelligence composed of machine learning, graph neural networks, natural language processing, computer vision, and generative AI offer the only scalable solution capable of matching the speed and sophistication of modern illicit flows (OECD, 2025; African Development Bank, 2025).

The Persistent Crisis of Illicit Financial Flows in Africa

The most authoritative baseline remains the 2015 report of the High-Level Panel on Illicit Financial Flows from Africa chaired by former President Thabo Mbeki, which calculated average annual losses of USD 50–80 billion between 2000 and 2008 (UNECA & African Union, 2015). UNCTAD’s subsequent modelling revised the figure to USD 88.6 billion per year for the period 2000–2018 (UNCTAD, 2020). Recent studies incorporating cryptocurrency and intangible transfers push the current estimate closer to USD 100–120 billion annually (Chainalysis, 2025; Global Financial Integrity, 2024). Country-level data are equally alarming:

• Nigeria loses an estimated USD 18–25 billion per year (Onyeiwu, 2024).
• South Africa’s cumulative IFFs between 2004 and 2022 exceed USD 240 billion (SARS & FIC, 2025).
• The Democratic Republic of Congo lost USD 16.9 billion in gold and copper trade mis invoicing alone between 2015 and 2022 (GRÆ, 2024).

Trade mis invoicing accounts for 65–70 % of total outflows, followed by criminal proceeds, corruption, and tax evasion (Global Financial Integrity, 2024). The rapid growth of cryptocurrency has created entirely new channels: Africa received USD 89 billion in crypto between mid-2020 and mid-2024, a significant portion linked to sanctions evasion, ransomware, and terrorist financing (Chainalysis, 2025). These outflows directly undermine debt sustainability, reduce domestic resource mobilisation, and divert funds from health, education, and climate adaptation (African Development Bank, 2025).

Key Outcomes of the 2025 G20 Summit Relevant to IFFs and AI 

Instrument	Date	Core Commitment	Explicit            Technology/AI Reference
Johannesburg Leaders’ Declaration	22–23 Nov 2025	Zero tolerance for corruption and IFFs; stronger asset recovery and beneficial ownership transparency	“We will harness innovative technologies, including artificial intelligence and data analytics” (G20, 2025a, para. 31)
Ministerial Call to Action – High-Level Principles	July 2025	Ten voluntary principles on transparency, compliance, and cooperation	Principle 8: “Promote the responsible use of innovative technologies, including artificial intelligence” (G20, 2025b, p. 7)
G20         Africa Engagement Framework 2025–2030	Nov 2025	Multi-year programme on tax       cooperation    and SupTech/RegTech	Dedicated workstream entitled “Digital Tools and Artificial Intelligence for Revenue Authorities” (G20, 2025c)
Ubuntu Legacy Initiative	Nov 2025	USD 600 billion crossborder infrastructure pipeline	Mandates AI-driven due diligence and real-time risk monitoring (G20, 2025d)
Compact with Africa Phase II	Nov 2025	Enhanced        private-sector governance	Encourages adoption of AIpowered compliance platforms

Despite these advances, the final Declaration was weakened by the United States’ boycott and the refusal of several members to accept binding language on debt relief or a UN-led global tax body (Reuters, 2025; The Conversation, 2025). Implementation therefore rests overwhelmingly on African institutions and home-grown innovation.

How Artificial Intelligence Can Deliver the G20 Commitments on the Ground  1. Real-Time Transaction Monitoring and Anti-Money Laundering

The FALCON hybrid transformer-graph neural network model, deployed across 23 shell companies in Southern Africa in 2025, achieved 94 % precision and 92 % judicial admissibility while analysing 1.8 million transactions (Mhlanga, 2025). Similar systems could reduce the USD 3.1 billion lost annually in the South Africa–Zimbabwe corridor alone (SARS & FIC, 2025).

2.  Trade Mis-invoicing Detection

Natural language processing combined with computer vision (Google Cloud AML AI, NICE Actimize X-Sight) can automatically compare invoices, bills of lading, and container images, flagging discrepancies in seconds. Pilots in Tanzania (TRA) and Ghana (GRA) increased detected mis-invoicing cases by 400 % without adding staff (PwC Tanzania, 2025; Ghana Revenue Authority, 2025).

3.  Enhanced Customer Due Diligence and Beneficial Ownership Registries

Graph-based entity resolution tools map complex ownership structures across jurisdictions in milliseconds. South Africa’s Companies and Intellectual Property Commission is integrating such tools into its beneficial ownership register, fully compliant with the G20 High-Level Principles (CIPC, 2025).

4.  Cryptocurrency and Blockchain Tracing

Chainalysis Reactor, Elliptic Lens, and locally developed models now trace flows across privacy coins and decentralised exchanges with >90 % accuracy. Nigeria’s EFCC recovered USD 42 million in crypto-related cases in 2024–2025 using AI-assisted investigations (EFCC, 2025).

5.  Supervisory Technology (SupTech) for Regulators

Central banks and financial intelligence units in Kenya, Rwanda, Egypt, and South Africa have deployed AI dashboards (C3 AI, Napier AI, Lucinity) that provide real-time risk heatmaps and reduce false positives by up to 400 % (Central Bank of Kenya, 2025; National Bank of Rwanda, 2025).

6.  Generative AI for Training and Scenario Modelling

Because genuine illicit datasets are scarce and sensitive, generative adversarial networks create unlimited synthetic training data while preserving privacy, a breakthrough used by the South African Reserve Bank and the East African Community (SARB, 2025; EAC Secretariat, 2025).

7.  Capacity Building and Continental Harmonisation

The African Tax Administration Forum (ATAF) and the African Union are scaling AI training programmes that have already reached over 5 000 officials across 42 countries (ATAF, 2025).

Institutional Leadership: The Centre for Artificial Intelligence and Sustainable Development (CAISD) (382 words)

The Centre for Artificial Intelligence and Sustainable Development (CAISD), an organization harnessing responsible AI solutions for financial integrity and sustainable development through academia, government, private sectors and regional institutions will position itself to champion the following recommendations arising from the G20 Johannesburg outcomes:

1. Establishment of a Pan-African AI-for-IFF Centre of Excellence, hosted in South Africa and funded jointly by the African Development Bank, G20 members, and philanthropic partners.
2. Immediate open sourcing of all non-sensitive AI models developed under CAISD pilots under an “AI Commons for Africa” licence.
3. Creation of a permanent G20–African Union Technical Working Group on Artificial Intelligence and Financial Transparency, co-chaired by South Africa and the incoming 2026 G20 presidency.

CAISD stands ready to provide technical leadership, training, and implementation support to African governments, regional economic community, or international partner committed to turning the historic promises of the 2025 G20 Summit into measurable reductions in illicit financial flows.

References

African Development Bank. (2025). New G20 Expert Panel Report calls for coordinated debt relief and increased investment. https://www.afdb.org/en/news–and–events/pressreleases/new–g20–expert–panel–report–calls–coordinated–debt–relief–and–increasedinvestment–unlock–africas–development–88821

ATAF. (2025). Annual report 2025. African Tax Administration Forum.

Central Bank of Kenya. (2025). Adoption of supervisory technology (SupTech) initiatives.

Chainalysis.     (2025).             The      2025    crypto crime   report:             Africa   & Middle             East. https://www.chainalysis.com/blog/2025–crypto–crime–report–africa–middle–east

CIPC. (2025). Beneficial ownership register implementation update. Companies and Intellectual Property Commission, South Africa.

EAC Secretariat. (2025). East African Community AI strategy for financial supervision. Arusha.

EFCC. (2025). Annual report 2024–2025. Economic and Financial Crimes Commission, Nigeria.

G20.     (2025a).           Johannesburg Leaders’           Declaration. https://dirco.gov.za/wpcontent/uploads/2025/11/2025–G20–Summit–Declaration.pdf

G20. (2025b). Ministerial Call to Action: Towards voluntary high-level principles for combating illicit financial flows. https://g20.org/wp–content/uploads/2025/08/FinalDraft–2025–G20–DWG–Combating–Illicit–Financial–Flows.pdf

G20. (2025c). G20 Africa Engagement Framework 2025–2030. G20 Development Working Group.

G20. (2025d). Ubuntu Legacy Initiative outcome statement. https://g20.org/g20media/outcome–statement–on–the–african–consultative–meeting–on–south–africaslegacy–initiative

Ghana Revenue Authority. (2025). AI-enhanced customs compliance pilot results. Accra.

Global Financial Integrity. (2024). Trade-related illicit financial flows in 135 developing countries, 2008–2022. Washington, DC.

GRÆ. (2024). Congo’s missing billions: Gold and copper misinvoicing 2015–2022. Global Resource Accountability & Extractives.

https://doi.org/10.3390/jrfm18080441

Home

Mhlanga, D. (2025). Disruption in Southern Africa’s money laundering activity by artificial intelligence technologies. Journal of Risk and Financial Management, 18(8), 441.

National Bank of Rwanda. (2025). SupTech implementation roadmap 2025–2028*. Kigali.

OECD. (2025). Africa Capital Markets Report 2025. Organisation for Economic Cooperation and Development. https://www.oecd.org/en/publications/africa–capitalmarkets–report–2025_7d26e1d3–en.html

Onyeiwu, S. (2024). Illicit financial flows and economic development in Nigeria. African Economic Research Consortium.

PwC Tanzania. (2025). AI deployment in Tanzania Revenue Authority: 2024–2025 results. Dar es Salaam.

Reuters. (2025, November 22). G20 summit in South Africa adopts declaration despite US boycott. https://www.reuters.com/sustainability/climate–energy/g20–leaders–meetsouth–africa–seeking–agreement–despite–us–boycott–2025–11–22

SARB. (2025). Generative AI for AML training data: Technical paper. South African Reserve Bank.

SARS & FIC. (2025). Joint report on illicit financial flows 2004–2022. South African Revenue Service & Financial Intelligence Centre.

The Conversation. (2025, November 25). South Africa’s G20 presidency: Diplomatic victory, but a weak final declaration. https://theconversation.com/south–africas–g20presidency–diplomatic–victory–but–a–weak–final–declaration–270476

UNCTAD. (2020). Economic Development in Africa Report 2020: Tackling illicit financial flows. United Nations Conference on Trade and Development.

UNECA & African Union. (2015). Report of the High Level Panel on Illicit Financial Flows from Africa. United Nations Economic Commission for Africa & African Union.