Ethical AI Policy
Preamble
Africa is not waiting for the world to decide what artificial intelligence means for the continent. Africa is building its own answer.
afrAIca (PTY) Ltd was established to serve that imperative: to place sovereign, ethical, and human-centred AI at the heart of African organisational transformation.
This Ethical AI Policy is not a compliance exercise. It is a declaration of how we operate, why we operate that way, and what we will not do. It governs every engagement we undertake, every recommendation we make, and every line of code we help write or deploy.
Our clients trust us because we are independent. They rely on us because we put their interests, their people, and their nations first. This policy makes that commitment concrete.
Section 1: Purpose and Scope
1.1 Purpose
This policy establishes the ethical principles, operational boundaries, and accountability standards that govern all AI-related activities conducted by afrAIca (PTY) Ltd. It applies without exception to our advisory services, development projects, implementation engagements, research outputs, and partnership activities.
1.2 Scope
This policy applies to:
- All employees, contractors, associates, and partners of afrAIca (PTY) Ltd
- All client engagements and project deliverables
- All AI systems designed, evaluated, recommended, or deployed under afrAIca guidance
- All use of AI tools within afrAIca internal operations
- All research, publications, and thought leadership produced under the afrAIca brand
1.3 Policy Owner
This policy is owned by the Office of the Chief Executive Officer and is subject to annual review or immediate revision in response to material changes in the regulatory, technological, or social environment.
Section 2: Governing Principles
afrAIca’s approach to AI ethics is grounded in eleven interdependent principles. They are not hierarchical. They are holistic. A failure in any one of them is a failure of the whole.
AI adoption in Africa must serve African interests. We advocate for and implement solutions that preserve human capital, algorithmic independence, data residency, and infrastructure control within African institutions and territories. We will not recommend solutions that erode an organisation’s or nation’s strategic autonomy over its own data and intelligence.
AI must create genuine organisational and societal progress. We measure success by the capability our clients build, not by the technology they purchase. Every engagement must leave the client more capable, more independent, and better positioned than when we arrived.
We accept accountability for the quality, accuracy, and consequence of our recommendations. We do not hide behind contractual disclaimers. If we advise it, we stand behind it. Responsibility extends to downstream effects on society, not only to the immediate client.
We will not engage with AI applications designed to deceive, manipulate, surveil without consent, or cause harm. Ethics is applied before a project begins, not appended as an afterthought. We conduct ethical impact assessments as a standard component of our engagement methodology.
Bias in AI systems reflects the assumptions of those who build them. We actively audit training data, model outputs, and assessment criteria for bias across race, gender, language, geography, and socioeconomic status. African languages, cultures, and contexts receive equal rigour, not afterthought accommodations.
No two organisations are alike. Our recommendations are derived from the client’s own context, objectives, and constraints, not from predetermined solutions or vendor preferences. We listen before we advise. We assess before we recommend. We ask before we assume.
We operate within an African philosophical tradition that recognises the person through community: Umuntu ngumuntu ngabantu. AI must strengthen the social fabric, not dismantle it. Solutions that isolate, exclude, or concentrate power in the hands of the few are contrary to our purpose. Technology must serve the many.
We are transparent about our methods, our limitations, and our reasoning. We do not create dependency through opacity. Clients receive clear explanations of how AI systems work, what their limitations are, and what risks they carry. We publish our positions and engage openly with critique.
Data about people belongs to those people. We apply privacy-by-design principles to every engagement. We align with the Protection of Personal Information Act (POPIA, Act 4 of 2013), the OECD AI Principles, and emerging African data governance standards. Cross-border data flows are assessed, disclosed, and managed with rigour.
We tell clients what they need to hear, not what they want to hear. We disclose conflicts of interest. We acknowledge uncertainty. We do not overstate AI capabilities or our own. If a proposed AI application is not fit for purpose, we say so, regardless of the commercial consequence.
AI augments human capability. It does not replace human dignity or human judgement. In every engagement, we prioritise the wellbeing, agency, and development of the people affected by AI systems. Technology serves people. Never the reverse.
Section 3: What We Will Not Do
afrAIca does not engage in AI projects where the primary or stated client objective is to reduce headcount. We believe AI should expand human capability and create new opportunities. We will not design, recommend, or implement AI systems whose primary purpose is workforce displacement.
We do not work on surveillance systems designed to monitor individuals without their knowledge or consent.
We do not build or recommend systems designed to manipulate human behaviour, exploit psychological vulnerabilities, or undermine individual autonomy.
We do not engage with projects that concentrate data or decision-making power in ways that undermine democratic accountability or national sovereignty.
Section 4: AI Readiness Assessment Standards
Our AI Readiness Assessments are conducted against five dimensions: Infrastructure and Connectivity, Data Availability and Quality, AI Talent and Skills, Policy and Regulation, and AI Adoption and Investment.
- Assessments are evidence-based, client-specific, and free from vendor influence.
- Findings are presented honestly, including gaps and risks, not only opportunities.
- Recommendations are prioritised by impact and feasibility within the client’s context.
Section 5: Responsible Deployment Standards
No AI system is recommended for deployment without:
- A defined use case
- Measurable success criteria
- A human oversight mechanism
- A data governance plan
- An exit or remediation path if the system underperforms or causes harm
These are not optional additions. They are preconditions for engagement.
Section 6: Human Oversight and Accountability
AI does not make decisions. People make decisions. AI informs them. In any AI application affecting individual rights, livelihoods, health, or access to services, we require a human review mechanism before consequential action is taken.
We recommend that every AI system deployed within a client organisation is:
- Governed by a named accountable person
- Documented in the client’s AI governance framework
- Subject to regular review
- Embedded within the organisation’s broader risk management and King V governance obligations
Section 7: Fairness and Inclusion
Before recommending or deploying any AI model in a high-stakes context, we conduct or require evidence of bias assessment across protected characteristics including race, gender, age, disability status, language, and geography.
AI systems must be accessible and functional for the populations they are intended to serve. We will not advise on deployments whose primary effect is to concentrate economic advantage within a narrow group while externalising costs onto marginalised communities.
Section 8: Privacy by Design
Privacy is not a control applied after a system is built. It is a design requirement from the first conversation. Our engagements apply:
- Data minimisation
- Purpose limitation
- Storage limitation
- Security by default
- Consent and transparency
- Cross-border transfer assessment under POPIA section 72
- Heightened controls for special category data
Section 9: Environmental and Social Responsibility
Where technically feasible, we recommend energy-efficient infrastructure and African renewable energy sources for AI compute. For deployments with significant societal reach, we recommend that clients conduct social impact assessments before deployment and establish ongoing monitoring mechanisms.
Section 10: Compliance Framework Alignment
afrAIca’s ethical standards are aligned with:
- POPIA Act 4 of 2013
- CSIR AI Maturity Assessment Framework
- OECD AI Principles
- King V Report on Corporate Governance
- ISO/IEC 42001 AI Management Systems
- African Union AI Policy Framework
- The SA National Development Plan
Section 11: Governance and Enforcement
The Chief Executive Officer holds ultimate accountability for this policy. No commercial consideration overrides a commitment contained in this policy. Where a client request conflicts with this policy, the project lead is obligated to escalate to the CEO before proceeding.
Concerns may be raised via progress@afraica.co.za. This policy is reviewed annually as a minimum.
Section 12: Our Commitment
We close with what we began with. AI is not neutral. It carries the values, assumptions, and intentions of those who build it. afrAIca chooses to build with intention.
We choose sovereignty over dependency. We choose honesty over convenience. We choose people over efficiency. We choose Africa.
This policy is our public commitment to every client, every partner, and every person whose life is touched by the AI systems we help to shape. We invite scrutiny. We welcome accountability. And we hold ourselves to this standard without exception.
Signed and Adopted
Chris Coetzee
Founder and Chief Executive Officer
afrAIca (PTY) Ltd
Date: November 2025
Policy Ref: afrAIca-POL-ETH-001