Ethical AI Examples: How to Embed Ethics in Your AI Projects
Embedding ethics into an AI project means asking three questions before deployment: who can be harmed by this algorithm, who is accountable for it, and how is it controlled. Concrete examples range from the UNESCO AI ethics framework to anti-discrimination guardrails in recruitment and the transparency requirements of the EU AI Act.
Why Ethics Is Not Optional
Morocco wants to move from AI potential to economic impact. That is the signal from this week’s news, from the Casablanca CEO Forum to government statements on digital sovereignty. But moving to impact without an ethical framework means building on sand.
A recruitment algorithm that systematically filters out female profiles or candidates from certain regions does not generate value. It creates legal risk, reputational risk, and operational risk. This is not philosophy. It is risk management.
Companies that deployed AI tools without solid AI governance have paid the price. Amazon shut down a CV screening tool in 2018 because it penalized female candidates. The tool had learned from historically biased data. Two years of development wasted, and devastating press coverage.
The Three Pillars of Ethical AI in Business
1. Algorithmic Transparency
A leader must be able to answer this question: why did this algorithm make this decision? If no one on your team can answer, you have a governance problem, not a technical one.
The EU AI Act, whose adoption by the European Parliament marks a structural milestone for AI regulation, sets transparency requirements for high-risk systems. If you operate in Europe or process data from European residents from Casablanca or Dakar, these rules apply to you.
A concrete example: a Moroccan bank using an AI-based credit scoring system must be able to explain to a customer why their application was rejected. Not “the algorithm decided.” An articulated, verifiable, contestable reason.
2. Non-Discrimination
This is the most sensitive use case in human resources. As I explained in my analysis of AI CV screening tools, automated sorting tools reproduce the biases present in their training data.
UNESCO’s Recommendation on the Ethics of AI, adopted by 193 states in 2021, is clear: AI systems must be regularly audited for discriminatory bias. Not once at deployment. Regularly.
For a CHRO in Morocco or Belgium, what does that mean in practice? Your recruitment tool must be tested on fictitious profiles that are identical except for gender, origin, or age. If scores diverge, you have a bias. You correct it before someone else discovers it for you.
3. Personal Data Protection
AI runs on data. In the African context, data sovereignty is a central issue. Where is your candidates’ or clients’ data stored? Who has access? Under which jurisdiction?
In Morocco, the Commission nationale de contrôle de la protection des données à caractère personnel (CNDP) oversees these practices. In Belgium and France, it is the GDPR. Both require explicit consent, data minimization, and the right to erasure.
An example of ethical AI on this point: a BPO company in Casablanca processing data from European clients must apply GDPR standards, even operating from Morocco. This is not an administrative constraint. It is a condition of access to the European market.
I built a 6-dimension diagnostic framework to assess the ethical maturity of an AI project before deployment. Download the Board Pack AI 2026 to use it directly with your team.
What Companies That Get It Right Actually Do
They do not treat ethics as an annual audit. They embed it in the development cycle, the same way they handle cybersecurity.
In practice, it looks like this: an ethics review committee that validates each new use case before deployment, documented evaluation criteria, and a named accountable person. Not a cosmetic committee. A body with the power to block.
The African Development Bank (AfDB) and the UNDP are accelerating the AI 10 Billion initiative, funded at one billion dollars per year, to deploy AI at scale across the continent. This kind of initiative can only produce lasting results if funded projects integrate guardrails from the design stage. African leaders positioning their organizations for this type of funding need to have their house in order on this topic.
For a deeper look at the concrete risks associated with AI, my analysis of the drawbacks of artificial intelligence covers the foundations every leader should know before deploying.
The Question Nobody Asks in the Boardroom
If the algorithm is wrong and someone suffers the consequences, who takes responsibility?
Not “who signed the purchase order.” Who is responsible and accountable, in the operational sense, if a candidate is wrongly rejected, a loan is unjustly denied, or an HR decision is challenged in court?
If you do not have a clear answer to that question, you do not have AI governance. You have an unattended tool.
If you are a CEO or CHRO and want to structure your approach before your next AI deployment, request a free diagnostic.
FAQ
What is a concrete example of ethical AI in business?
A bank that explains to each customer the reasons for an algorithm-generated credit refusal, with verifiable criteria and a right of appeal, is an example of applied ethical AI. The tool makes the decision, but a human remains accountable and the process is transparent.
Does the EU AI Act apply to Moroccan companies?
Yes, if they process data from European residents or offer services on the European market. The EU AI Act sets reinforced requirements for AI systems used in sensitive contexts such as recruitment or credit assessment. If your business touches the European market, these rules apply to you directly.
How do you detect bias in a recruitment algorithm?
By testing the tool with fictitious profiles that are identical except for gender, first name, or geographic origin. If evaluation scores diverge systematically, the bias is confirmed. This audit must be repeated at regular intervals, not only at deployment.
What is the difference between AI conformity and AI ethics?
Conformity means following existing rules. Ethics means anticipating impacts on people before the rules exist. A company can be GDPR-compliant and still deploy a discriminatory algorithm. The two approaches are complementary, not interchangeable.