South-South Cooperation in AI Policymaking: Developing a Collaboration Roadmap

Abstract

The panel convened members of the Africa‑Asia AI Policymaker Network to map out concrete South‑South collaboration mechanisms for AI governance. After a brief network overview, the discussion moved through a draft “road‑map” that highlighted six priority mechanisms (regulatory harmonisation, joint skills programmes, shared infrastructure & data, cross‑border sandboxes, collaborative audit, and peer‑exchange). Each of the seven country representatives then presented the status of AI policy in their nation, underscoring challenges, successes, and concrete ideas for joint action. The session concluded with a synthesis of shared priorities—especially infrastructure pooling, data‑exchange frameworks, and capacity‑building programmes—followed by a short Q&A and closing remarks.

Detailed Summary

• Wolfger Bungarten (BMZ) opened the session, thanking the Global Center on AI Governance, the Human Sciences Research Council (HSRC), and the host nations.
• Rachel Adams welcomed the audience, explained the purpose of the Africa‑Asia AI Policymaker Network (AAA‑PN), and highlighted its founding in 2022 with seven member countries (Ghana, India, Indonesia, Kenya, Rwanda, South Africa, Uganda). She noted BMZ’s ongoing funding for organisational structures and capacity‑building exchanges.
• She emphasized the network’s twin achievements: (i) an AI policy playbook offering practical guidance to regulators, and (ii) a growing cross‑border community that fosters “fair, transparent, inclusive, safe and responsible” AI.

2. Draft Collaboration Roadmap (Presented by Dr Odilele Ayodele)

• The draft discussion document (available via QR code) outlines six core mechanisms for South‑South cooperation:

  1. Regulatory harmonisation & mutual recognition for AI‑enabled services and devices.
  2. Joint AI skills programmes & talent mobility, including shared curricula and exchange scholarships.
  3. Shared infrastructure, data exchanges & collaborative development models (e.g., compute clouds, test‑beds).
  4. Cross‑border regulatory sandboxes & experimentation frameworks that allow innovators to trial AI under joint supervision.
  5. Collaborative AI audit & monitoring mechanisms, including joint incident‑response centres.
  6. Policy peer‑exchange, advocacy & multi‑stakeholder participation (annual forums, joint statements).

• She stressed that these mechanisms aim to move from “normative discussion” to tangible practice, ensuring AI benefits and risks are equitably distributed across the Global South.

3. Country‑Specific Presentations

3.1 Kenya – Richard Kiari

• Kenya’s AI journey began 2022 with a stakeholder‑driven AI strategy (supported by GIZ).
• Key challenges: lack of local infrastructure, dependence on foreign AI solutions, and fragmented data governance.
• Priority actions:
  • Develop regional compute infrastructure (shared with other Southern African nations).
  • Create a cross‑border data‑governance framework for agriculture, health, weather, and other sectoral data.
  • Promote ethical, responsible, and inclusive AI to guard against disinformation and security threats.
  • Establish forums for best‑practice sharing and a minimum cross‑border data‑transfer protocol.

3.2 Indonesia – Kautsarina Adam

• India’s AI Mission (2024), not Indonesia’s, but Kautsarina referenced Indonesia’s AI ecosystem within the broader mission context (the network includes both).
• Six‑pillar budget of ₹10,000 crore (≈ US $1.2 bn) covering:
  1. Infrastructure – subsidised GPU cloud access (≈ US $1 per hour).
  2. Sovereign AI models – development of Indian‑language LLMs (“Bharat Jan”, “Sarvam AI”).
  3. Data platforms – > 2,000 public datasets for startups and academia.
  4. AI Safety Institute – governance, ethics, bias mitigation.
  5. Skilling – nationwide AI literacy programmes.
  6. Economic development roadmap – AI for GDP growth (target $30‑40 trillion by 2047).
• Highlighted Digital Public Infrastructure (DPI) (e.g., UPI‑style payments) as a model for scalable, low‑cost AI services.
• Proposed South‑South exchanges on DPI design, data‑exchange know‑how, and AI‑enabled soil‑analysis pilots (already implemented in Telangana).

3.3 Rwanda – Cyprien Nshimiyimana

• Rwanda defined its National AI Policy to position the country as a global AI research hub.
• Economic impact study (2022) estimated AI would contribute ≈ 6 % of GDP (pre‑ChatGPT era), likely higher now.
• Priority sectors: health, agriculture, education.
• Enablers: digital skills, AI literacy, data & compute infrastructure; safeguards: ethical AI guidelines (2023).
• South‑South collaborations:
  • AI Scaling Hub (partnered with Gates Foundation) to help pilots scale via compute resources and policy support.
  • AI Playbook for Small States (co‑developed with Singapore, 2024).
  • AI Fellowship (with UAE & Singapore) – over 100 k civil servants certified via Dubai AI Centre.
• Calls for greater investment in knowledge‑sharing nodes and continued AI‑summit participation.

3.4 India – Rama Devi Lanka

• Introduced India’s AI Mission (2024) with 10,000 crore INR budget and six pillars (infrastructure, models, data, safety, skilling, economic development).
• Infrastructure: 10,000 subsidised GPUs for researchers and start‑ups.
• Sovereign models: development of LLMs in 22 Indian languages (e.g., “Bharat Jan”).
• Data: national data platform with ~2,000 datasets made accessible.
• AI Safety Institute: focuses on governance, ethics, bias.
• Skilling: large‑scale AI‑literacy programmes for students, teachers, and industry.
• Economic development: AI‑for‑GDP roadmap (targeting $30‑40 trillion by 2047) and AI‑for‑informal‑workforce initiatives (credentialing, tool provision).
• South‑South cooperation:
  • Offer data‑exchange templates (drawn from Telangana’s DPI) to Kenya, Rwanda, Indonesia.
  • Share DPI expertise to reduce implementation time for other countries.
  • Provide a playbook repository of AI use‑cases for agriculture, education, etc.
• Cited a concrete soil‑analysis project in Telangana that originated from a Kenya‑India discussion.

3.5 Uganda – Irene Karungi Sekitoleko

• Uganda produced a national AI landscape assessment feeding directly into a draft AI strategy (aligned with the 2024 Digital Transformation Roadmap).
• Enacted a personal data‑protection & privacy law to underpin trusted AI.
• Established a multi‑sectoral AI Task Force (2025) to coordinate policy, ethics, and implementation.
• Capacity building: training for civil servants and universities; establishment of AI labs.
• Aligning with global standards via UNESCO‑supported ethical‑AI readiness assessment and participation in the AI supervisory network.
• Hosted the Global AI Innovation Movement & Evolution Conference in Kampala.
• Future cooperation goals:
  • Leverage Digital Public Infrastructure for secure digital ID and AI services.
  • Harmonise governance with regional bodies (East African Community AI strategy).
  • Co‑develop guidelines for generative AI.
  • Joint skills, research, and compute‑capacity programmes (exchange programs, shared datasets).

3.6 South Africa – Siphokazi Novukuza

• AI policy process started 2024, with public consultations and sector‑specific clusters; Parliament vote expected March 2025.
• Built upon existing Protection of Personal Information Act and Cyber‑Crimes Act to integrate AI‑specific cybersecurity protocols.
• Established four AI hubs in partnership with universities, funded by a R300 million Microsoft commitment for AI infrastructure.
• South‑South cooperation focus:
  • Peer regulatory exchange to design innovation‑friendly rules and sandbox environments.
  • Compute partnerships (e.g., with Kenya’s G42, UAE).
  • Common data‑sharing principles drawn from COVID‑19 lessons for health surveillance and epidemic response.

3.7 Ghana – Maxwell Ababio

• Ghana’s AI journey began 2022 with a multistakeholder drafting process now pending cabinet approval.
• Implemented a workshop with cabinet ministers to embed AI policy awareness at the highest level.
• Introduced “AI focal persons” in each ministry (mirroring data‑protection supervisors) to oversee AI adoption and avoid “copy‑paste” solutions.
• Ran capacity‑building training on prompt‑engineering for senior officials.
• Benchmarking with Rwanda’s implementation was a key learning activity.
• Cross‑border data framework developed at the continental level to foster trust and enable data sharing.

4. Cross‑Cutting Themes & Emerging Consensus

Theme	Common Points Across Countries
Infrastructure & Compute	Need for shared cloud/compute resources; Kenya, Rwanda, South Africa, and Ghana all mention pooling or joint procurement.
Data Governance & Sharing	Kenya, Uganda, and Ghana stress a common data‑exchange framework; India offers DPI templates; South Africa cites COVID‑19 data‑sharing lessons.
Regulatory Harmonisation	All participants endorse mutual‑recognition of AI standards; Kenya, Indonesia (via India), and Rwanda emphasise sandbox co‑design.
Skills & Talent	India’s massive GPU‑subsidy & skilling drive, Rwanda’s AI fellowship, Ghana’s prompt‑engineering workshops, and Uganda’s task‑force capacity‑building.
Policy Peer‑Exchange	The AAA‑PN playbook, regular network meetings (PumaNet), and an annual AI summit are highlighted as knowledge‑transfer mechanisms.
Ethics & Trust	Every speaker refers to ethics bodies, AI safety institutes, or trust‑building data frameworks as pre‑condition for collaboration.
South‑South Funding Models	Rwanda‑Singapore playbook, India‑UAE AI fellowship, and South Africa’s Microsoft partnership illustrate diversified financing.

5. Q&A Highlights

• Question on legal enforceability of cross‑border sandboxes – Richard Kiari noted that Kenyan regulators are drafting a “sandbox charter” aligned with Kenya’s Data Protection Act and would welcome harmonised templates from the network.
• Query about AI literacy for informal workers – Rama Devi Lanka described a credential‑system pilot in Telangana that issues digital badges for carpenters and plumbers, suggesting it as a model for other nations.
• Concern about data‑sovereignty – Irene Karungi emphasized that Uganda’s personal‑data law serves as a template for ensuring that shared datasets respect national sovereignty.

6. Closing Remarks

• Rachel Adams thanked all panelists, highlighted the six‑point roadmap as the session’s concrete output, and called on participants to review the draft document, provide feedback, and commit to at least one of the mechanisms before the next AAA‑PN meeting.
• Wolfger Bungarten announced a follow‑up virtual workshop (late May) to pilot a cross‑border data‑exchange prototype among Kenya, Rwanda, and Uganda, funded by BMZ.

Key Takeaways

• The Africa‑Asia AI Policymaker Network (AAA‑PN) now has a draft road‑map centred on six cooperative mechanisms: regulatory harmonisation, joint skills programmes, shared infrastructure & data, cross‑border sandboxes, collaborative audit, and peer‑exchange.
• Infrastructure pooling (compute clouds, GPU subsidies, AI hubs) is a shared priority; India’s GPU‑subsidy scheme and South Africa’s Microsoft partnership are cited as replicable models.
• Data governance emerges as the “trust” bottleneck; Kenya, Uganda, and Ghana are developing regional data‑exchange frameworks, while India offers DPI templates that can cut implementation time.
• Regulatory sandboxes are being piloted in Kenya and South Africa; harmonised sandbox charters are proposed to accelerate responsible innovation across borders.
• Skills development is tackled through large‑scale national programmes (India’s AI Mission), fellowships (Rwanda‑UAE‑Singapore), and ministry‑level AI focal persons (Ghana).
• Ethical oversight (AI safety institutes, UNESCO‑backed readiness assessments) is now embedded in national strategies across the board.
• South‑South funding models (Gates Foundation, Microsoft, BMZ) demonstrate that external development partners can catalyse collaborative projects without creating donor dependency.
• Concrete next steps: review and comment on the draft discussion document, pilot a cross‑border data‑exchange prototype (Kenya–Rwanda–Uganda), and submit national action plans aligned with the six mechanisms before the next network summit.

---

Prepared from the verbatim transcript of the panel “South‑South Cooperation in AI Policymaking: Developing a Collaboration Roadmap” held in Delhi.

See Also:

• thriving-with-ai-human-potential-skills-and-opportunity
• governing-safe-and-responsible-ai-within-digital-public-infrastructure
• ai-beyond-moonshots-a-playbook-for-many
• data-sharing-infrastructures-for-ai-building-for-trust-purpose-and-public-values
• ai-innovation-and-collaboration-shaping-resilient-economies
• accelerating-indias-ai-growth-a-blueprint-for-indias-ai-success
• ai-for-democracy-reimagining-governance-in-the-age-of-intelligence