1. Project summary

AI is being deployed rapidly across Southern Africa and the wider Global South, yet the capacity to govern it safely has not kept pace. The UN High-level Advisory Body on AI found that 118 countries, mostly in the Global South, are party to none of the sampled international AI governance initiatives. Oxford Insights' 2025 Government AI Readiness Index ranks Sub-Saharan Africa last among global regions. At the centre of this gap is a scarcity of AI safety talent. Southern Africa and the wider Global South have very few trained professionals and dedicated organisations working on catastrophic and existential AI risk. This leaves governments without the expertise to assess risks, limits the region's voice in global governance, and exposes low-resource countries to harms from systems built elsewhere, with no local capacity to detect or respond to them. Because severe AI failures could cross borders, this is a regional and global vulnerability.

The Southern Africa AI Safety and Governance Initiative is a planned non-profit, to be registered in Malawi, established to build this talent and institutional capacity. Our model has three components. First, a global fellowship and volunteer programme will recruit experienced AI safety and catastrophic-risk experts, who will help build the organisation and train local talent identified through a fellowship programme. Second, trained local fellows will be placed in strategic government and private-sector institutions, each mentored by a global expert. There they will build the systems that reduce catastrophic risk, including risk-assessment procedures, oversight mechanisms, policies for the safe adoption of advanced AI, and preparedness for severe AI failures. Third, we will conduct original research on AI safety in Southern Africa

Our contribution to catastrophic and existential risk mitigation works through institutions rather than individuals alone. Embedded fellows will create standing safeguards that outlast any single placement, a growing regional network strengthens cooperation and reduces blind spots, and Global South perspectives enter international governance. Partnerships with established AI safety organisations will provide technical support and knowledge exchange.

We are requesting for a pre -seed fund amounting to $10,000 to assist in the registration, establishment and running of this newly established organisation.

What are this project's goals? How will you achieve them?

The overarching goal is to build the talent and institutional capacity Southern Africa needs to prevent AI-related harms and manage catastrophic and existential risks from advanced AI. We pursue it through five goals.

1. Establish a credible, operational organisation in Malawi
We will complete legal registration with the Registrar of Companies, the NGO Regulatory Authority and CONGOMA, set up governance structures, an office and a public website, and secure advisors, mentors and partner organisations. This gives the initiative the legal standing, visibility and support network needed to receive funds and attract expertise.

2. Build a pipeline of local AI safety talent
A global volunteer and expert fellowship will recruit experienced AI safety and catastrophic-risk professionals. In the first phase, they will train local talent identified through the fellowship programme and help build the organisation, addressing the regional scarcity of trained professionals at its source.

3. Embed fellows where they can reduce risk
A pilot cohort of five fellows, each paired with a global mentor, will be placed in government and other strategic institutions. There they will build lasting safeguards, including risk-assessment procedures, oversight mechanisms and policy for the safe adoption of advanced AI, expanding to ten fellows in year two once pilot lessons are applied.

4. Generate regional evidence on AI safety
We will conduct original research on AI safety in Southern Africa and convene policymakers, academics and civil society, beginning with Malawi's first AI safety conference. This fills an evidence gap and brings Global South perspectives into international governance.

5. Scale across Southern Africa on evidence
After the Malawi pilot, we will cascade the model to other Southern African countries, extending recruitment through trained fellows and partners. Embedded monitoring, evaluation and learning will track placements, institutional change and fellow outcomes, so expansion decisions rest on evidence. We will pursue follow-on funding and recruit staff throughout, so the work continues beyond the initial grant.

How will this funding be used?

The requested $10,000 will fund the foundational costs of establishing the initiative as a legal and operational organisation in Malawi. It will cover registration fees and legal fees for preparing and reviewing the documents submitted to the Registrar of Companies, the NGO Regulatory Authority and CONGOMA, so that the organisation is properly constituted and compliant. It will also fund a basic operational base: office rental, furniture, a computer and an internet connection. Finally, it will pay for the design, development and hosting of a website, giving the organisation the public presence that funders and partners expect. The grant does not cover fellowship or training costs. Once the organisation is registered and operating, we will pursue follow-on funding to launch the pilot fellowship.

Who is on your team? What's your track record on similar projects?

The initiative is currently founder-led. I am preparing the registration documentation and have drafted an expression of interest to recruit a co-founder with technical expertise in AI safety, specifically catastrophic and external risk. This person will help launch the organisation and make it operational within four months. In parallel, I am identifying advisors and expert volunteers, in Malawi and globally, who will provide technical guidance and mentorship. Because of funding constraints, I have deliberately kept the team lean at this pre-founding stage. We are prioritizing the one gap that matters most, deep technical AI safety expertise, and will recruit further staff once funding allows. My own role covers organisational set-up, programme design, grant management, and monitoring, evaluation and learning.
In terms of track my record I bring 16 years of experience in project management, with a specialization in monitoring and evaluation, across eight international NGOs, including Islamic Relief Worldwide, GOAL, Concern Worldwide and Project Concern International. I have served as Head of M&E on three USAID-funded projects and as Consortium M&E Technical Lead for a three-NGO consortium under the Start Network, which gave me direct experience of leading people and aligning priorities across organisations. I have also supervised teams of 10 to 15 staff and coordinated the recruitment, training and supervision of 20 field enumerators. I single-handedly drafted a research proposal that won a £15,000 Start Network UK Learning Grant, one of only two awarded globally, and delivered it through COVID-19 restrictions, submitting the final report on time. My full-scholarship International project management fellowship at the University of Catania in Italy in 2021, where I was selected among the top 30 candidates from 144 countries, gave me the tools to manage international projects successfully, and I apply them to ensure every programme I lead is designed for results and accountability. This combination of grant-winning, delivery under constraint and leadership at team and consortium level is the foundation on which I will build the initiative.

What are the most likely causes and outcomes if this project fails?

The following are the most likely causes of failure and their outcomes

1. Funding shortfall
Without funding for registration and early operations, the organisation could not complete legal registration, cover basic costs such as office space, or recruit and retain technical experts. The outcome would be a stalled initiative with no legal standing, limited credibility with partners, and no means of sustaining the work beyond the start-up phase. Early signs would be registration still incomplete by month four and no funder conversations moving past first meetings. We would respond by opening funder conversations from the first month, applying to several funders in parallel, and keeping costs to the essentials.

2. Insufficient founder time
If funding delays my transition from full-time employment , the project would receive evenings and weekends instead of 40 hours a week. Milestones would slip, partners and funders would see slow progress, and momentum would fade. This risk is tied directly to the funding risk. Availability of funding therefore makes it possible for full-time commitment of funders and experts.

3. Inability to secure technical expertise
Technical depth in catastrophic-risk work is scarce in the region. If we cannot recruit a co-founder, advisors and expert volunteers, the programme would lack the technical credibility to train fellows and win host institutions' confidence. The outcome would be a weaker programme and lower-quality fellow preparation. We would respond through the open expression of interest, partnerships with established AI safety organisations, and a volunteer expert fellowship that does not depend on salaries.

4. Registration delays
Registration involves the Registrar of Companies, the NGO Regulatory Authority and CONGOMA, so regulators' timelines are outside our control. Delays would prevent us from receiving funds or operating legally. We are preparing the documents in advance and budgeting for legal review to reduce this risk.

5. Reluctance of institutions to host fellows
Government and private institutions may hesitate to host unfamiliar fellows from a new organisation. The outcome would be trained fellows without placements and a dilution of impact. We would narrow the pilot to one or two partner institutions, begin placement conversations early, recruit fellows that are already working in the system and shift weight toward research and policy advice, which depends less on host institutions.

6. Dependence on a single founder
Until a co-founder and advisory structure are in place, too much rests on one person. We will mitigate this by recruiting the co-founder early, forming an advisory group and documenting governance from the start.

Overall outcome: The most likely failure is an organisation that is registered but unsustainable, or one that never fully launches. Southern Africa's AI safety talent gap would then persist, and the funder's investment would deliver only partial results. We have designed the plan so that these risks are monitored early and the scope is adjusted promptly, with the funder informed of setbacks and the steps taken to address them.
How much money have you raised in the last 12 months, and from where?

We have raised $0 to date. The Southern Africa AI Safety and Governance Initiative is at the founding stage: we are finalizing the documents for submission to the Registrar of Companies, and the organisation does not yet exist as a legal entity. Fundraising has begun, however. A pre-seed application has been submitted to Blue Dot Impact, and a decision is pending. If successful, it would support the establishment of the organisation and the piloting of the first fellowship programme. We are also applying to further funders, including this one, to cover registration and set-up costs.