When AI Fails: Building Accountability, Safety, and Trust in African Machine-Learning Systems

Project overview

AI Safety Nigeria (AISN), with the support of SPRITE+, convened a two-day Innovation Forum in Abuja titled “When AI Fails: Building Accountability, Safety, and Trust in African Machine-Learning Systems” on 18-19 August 2026. The hybrid, closed-door workshop brought together approximately 25 researchers, regulators, technologists, and civil-society practitioners; alongside additional 20 online participants — to move beyond discussion and directly co-author practical accountability tools for AI governance in Nigeria and across Africa.

Day 1 grounded participants in documented evidence of AI harms, including biased credit-scoring, especially gender-based bias in fintech systems, before moving into a facilitated session mapping the data, accountability, and governance gaps across African AI ecosystems. Day 2 built on this foundation with a technical session on AI control and the case for an African-specific AI harm taxonomy, a presentation on designing proportionate accountability tools, and further case simulation work on credit scoring and debt recovery, alongside case studies on facial recognition deployment in Zimbabwe and data sovereignty.

Key findings included the implications of the continued absence of a dedicated African AI Safety Institute, weak enforcement of existing data protection provisions, fragmented regulatory jurisdiction across Nigerian institutions, the need for

citizen-led interventions to address obstinate governance gaps, and the persistent underrepresentation of African contexts, languages, and populations in AI training data. Participants agreed that mitigations must be prioritised over redress mechanisms, while conceding that both strategies must complement the other; that gender and data sovereignty must be built into AI systems from the outset; and that governance cannot be left to the government alone.

The forum's two lasting outputs are the African AI Accountability and Trust Playbook, currently being drafted , and the Trusted AI for Africa Network, formally launched on Day 2 to sustain collaboration between researchers, regulators, industry, and civil society across the continent beyond the workshop itself.

Objectives

As set out in the original Expression of Interest, the forum's six objectives were to:

1. Examine documented cases of AI harm, bias, and erosion of trust in African settings; in fintech credit scoring, healthcare, public services, and emerging surveillance applications and to build a shared empirical foundation for the discussions.

   Achieved: covered in the Day 1 session on documented harms (credit-scoring bias, the Kano textile manufacturers' lawsuit, the Lagos Buy-Now-Pay-Later scandal, and the collapse of Babylon Health's diagnostic platform in Rwanda) and revisited via the CloudWalk facial recognition case study (Zimbabwe) on Day 2.

2. Identify and map structural gaps in Nigeria's and Africa's current accountability mechanisms.

   Achieved: delivered through the Day 1 “Mapping the Gaps in Africa's Governance” workshop (facilitated by Hanniel Jafaru), which examined the AI data, accountability, and governance gaps across national and continental frameworks. Hanniel structured this around three connected national gaps:

   a) The AI Data Gap (does the country have the right data to build, train, and evaluate AI?)

   b) The AI Accountability Gap (when AI causes harm, who answers for it?)

   c) The AI Governance Gap (institutions, laws, capacity, and coordination across the lifecycle), sitting alongside a broader AI infrastructure and skills gap (compute, connectivity, energy).

   It was grounded on current data: Nigeria ranks 70th of 195 countries on Oxford Insights' 2025 Government AI Readiness Index, and the African Union adopted its Continental AI Strategy in June 2024. Ten specific accountability gaps (AI audit, algorithmic transparency, explainability, human oversight, AI liability, bias testing, impact assessment, redress, incident reporting, and procurement/vendor accountability) and six governance breakdown points (policy, regulatory, institutional, capacity, coordination, and enforcement gaps) were mapped in detail.

3. Build a shared cross-sectoral understanding of ethical, technical, and policy responses, drawing on comparative lessons from the EU AI Act, the NIST AI Risk Management Framework, and the AU Continental AI Strategy.

   Achieved: delivered through the Day 2 technical session (Dr. Ayuba Tuodolo), which grounded participants in comparative governance models before extending into a technical discussion of loss of AI control and the case for an African-specific AI harm taxonomy ; a natural deepening of this objective in response to participant interest, rather than a departure from it.

4. Co-design practical accountability tools and governance recommendations proportionate to the African context.

   Achieved: delivered through Prof. James Agajo's technical presentation and workshop on “Designing Proportionate Accountability Tools” (detecting, documenting, explaining, and remedying harm). Prof. Agajo’s session introduced the ALARP principle (“As Low As Reasonably Practicable”, borrowed from aviation and nuclear safety) and a proportionality formula: accountability intensity in proportion to risk, impact, scale, and vulnerability mapped to four risk tiers. Eight concrete tools were presented (AI Accountability Impact Assessment, Data Quality & Bias Audit, Model Documentation & System Cards, Explainability & Decision Traceability, Human Oversight, Audit Trails & Evidence Logs, Incident Reporting & Redress, and Continuous Monitoring), worked through a live credit-scoring case study mapping ten risk areas to specific tools, practical solutions, and evidence/KPIs.

5. Forge lasting professional connections across research, government, and industry.

   Achieved: the room included representatives from civil society and industry (including CSEA, Accountability Lab, Paradigm Initiative, Women Rights Advancement and Protection Alternative, and Tech Hive Advisory) alongside academic and government participants (, and this cross-sector base was formalised through the Trusted AI for Africa Network launched on Day 2.

6. Produce at least two concrete proposals for future applied research or policy initiatives, ready for submission to funders within three months.

   Achieved: this work was carried out during the Day 2 “From Forum to Funders” session, with proposals identified for progression toward EPSRC, the Gates Foundation, the African Union, and the UK FCDO over the coming three months.

No changes were made to the original objectives; all six were delivered as planned.

Outcomes & outputs

Impact domains and stakeholders (per the original proposal's theory of change): the forum targeted impact across four stakeholder groups: academia, government/regulation, technology industry, and civil society; with the underlying logic that convening them jointly, rather than consulting them separately, would produce more durable governance outputs. All four groups were represented in the room across both days (e.g. Federal University Minna, and independent researchers; NITDA-adjacent policy discussion; fintech and tech-advisory practitioners; CSEA, Accountability Lab, Paradigm Initiative, and Women Rights advocates), matching the sectors identified at proposal stage.

Indicators and targets set at proposal stage vs. actual: (i) 25 participants across sectors achieved; (ii) two lasting outputs (Playbook + Network); both delivered; (iii) at least two fundable proposals within three months — identified during the forum and on track for submission; (iv) a SPRITE+ position paper strengthening African voice in global AI safety; evidence base secured at the forum, drafting to follow alongside the Playbook (see Future Work).

Social: discussions on gender bias in fintech credit-scoring, and the case for embedding gender considerations into AI systems and organisational culture from the outset, are feeding directly into the Playbook, with potential to improve financial-inclusion outcomes for women-led SMEs and rural borrowers.

Economic: the forum's focus on proportionate, resource-appropriate accountability tools — rather than direct transplantation of EU or US frameworks — aims to support innovation and investment in Nigeria's AI sector while reducing the economic harm documented in cases such as the Lagos Buy-Now-Pay-Later scandal and biased fintech underwriting.

Organisational: the Trusted AI for Africa Network creates a standing structure linking AI Safety Nigeria with Federal University of Technology Minna, NITDA, NDPC, NCAIR, and the Nigeria AI Collective, strengthening coordination across institutions that currently hold fragmented AI governance mandates — directly addressing the institutional fragmentation identified as the core problem in the original proposal.

Policy: the African AI Accountability and Trust Playbook is intended as a free, Africa-authored reference for policymakers and developers, directly addressing gaps identified in the unfinished NITDA Code of Practice and weak NDPA Section 37 enforcement. The Playbook draws on Prof. Agajo’s seven-step Accountability-by-Design framework (Identify → Assess → Control → Evidence → Monitor → Remedy → Learn), which explicitly operationalises NIST's Govern → Map → Measure → Manage approach, grounded in the current comparative data Nigeria's ranking of 70th of 195 countries on Oxford Insights' 2025 Government AI Readiness Index, and the African Union's Continental AI Strategy, adopted June 2024.

Unintended and indirect effects: two themes emerged that were not scoped in the original proposal — the case for an African-specific AI harm taxonomy (raised in the Day 2 technical session), and an emerging joint initiative on African-representative AI training datasets (raised by Hanniel Jafaru and Prof. James Agajo). Both are being carried forward as additional, unplanned outputs of the forum. The keynote by Andrew Ayobami additionally surfaced a multilingual safeguards mismatch and untested assumptions in some African governance mechanisms (e.g. open-weight models versus API-permission-based approaches) as under-researched risk areas, and named three sovereignty gaps affecting Africa's ability to address AI harm: infrastructure dependence, market leverage, and knowledge/information asymmetry. This converges with a design principle echoed independently by both technical facilitators: accountability tools must reflect local languages, informal economies, and connectivity constraints rather than importing controls that are impossible to implement locally.

Next step in AISN's impact-measurement workflow: findings from both days will be synthesised into a full impact narrative once the Playbook and Network charter are finalised, and reported through AISN's internal review before being shared with SPRITE+ Comms and re-branded for publication on the SPRITE+ website, consistent with the Measuring Impact Workflow shared with AISN at proposal stage.

Future work

• Completion and publication of the African AI Accountability and Trust Playbook within three weeks of the forum, incorporating all sections drafted during the workshop.

• Formalisation of the Trusted AI for Africa Network's founding charter with partner institutions (CAI University of Abuja, NITDA, NDPC, NCAIR, and the Nigeria AI Collective), building on its launch at the forum.

• Drafting and submission of the SPRITE+ position paper committed to at proposal stage, synthesising the forum's evidence base to strengthen African representation in international AI safety debates.

• Development of the African-representative AI dataset initiative raised between Hanniel Jafaru and Prof. James Agajo during the forum.

• Submission of the two fundable research and policy proposals identified during the forum to EPSRC, the Gates Foundation, the African Union, and the UK FCDO within three months.

• Completion of AISN's internal impact-measurement process (per its Measuring Impact Workflow): collecting data on Playbook adoption and Network activity against the indicators above, synthesising this into a full impact narrative, and sharing it with SPRITE+ Comms for re-branding and publication on the SPRITE+ website.

• AISN's Learn and Launch Series continues with its August edition on “AI Control & Corrigibility”, providing a further avenue for stakeholders to engage with themes raised at the forum.