Contents
The uses concerned
| Use | Main issue |
|---|---|
| Consumer credit scoring | Decision to grant or refuse, risk of discrimination |
| Life and health insurance pricing | Access to cover at a fair price |
| Fraud detection | Blocking of accounts or transactions |
| Banking chatbot | Accuracy of information on products and fees |
| Anti-money laundering | Suspicious activity reports and false positives |
What Law 09-08 says
Article 11 prohibits basing a decision with legal effects on a person solely on the result of automated processing intended to define their profile or assess certain aspects of their personality. The law provides for exceptions, notably where the decision is taken in the context of a contract and the person has been able to present their point of view.
In practice, a credit refusal based solely on an automated score, with no human involvement and no opportunity for the customer to put their case, exposes the bank. The rights of access and objection under Law 09-08 also mean being able to explain, in understandable terms, the logic of a score that has effects on a person.
What the AI Act says
Annex III of the AI Act classifies as high risk:
- systems intended to evaluate the creditworthiness of natural persons or establish their credit score, with the exception of those used to detect fraud;
- systems intended for risk assessment and pricing in relation to natural persons in life and health insurance.
These obligations apply from 2 December 2027. A Moroccan bank is affected when these systems produce output used in the European Union, for example through subsidiaries or activities involving customers who live there. Each situation must be assessed case by case.
Governance good practice
- An inventory of the models used in customer decisions, with an owner for each one.
- Human review of any significant adverse decision, with a channel for the customer to present their point of view.
- Bias testing on data representative of the Moroccan customer base, by region, age, sex and income type.
- Explainability: being able to tell the customer which factors weighed in the decision.
- Monitoring of model drift and periodic revalidation.
- Oversight of third-party models: documentation, test results, conditions of use.
- Integration into the internal control and risk management framework that Bank Al-Maghrib (Morocco's central bank) requires across all of a bank's activities.
See also The AI System Life Cycle: From Design to Retirement.
Further reading
- 4 min read
- 5 min read
- 4 min read
- 7 min read
Frequently asked questions
Can a bank refuse credit based solely on an automated score?
Article 11 of Law 09-08 prohibits basing a decision with legal effects solely on the result of automated profiling, subject to exceptions, notably in the context of a contract where the person has been able to present their point of view. Human review and a way for the customer to be heard are therefore strongly recommended.
Is AI-based fraud detection high risk under the AI Act?
No. Annex III explicitly excludes systems used to detect fraud from the high-risk category covering creditworthiness assessment.
When do the AI Act obligations for credit scoring apply?
From 2 December 2027, the date set by Regulation (EU) 2026/1744 for Annex III high-risk systems, and only where the AI Act applies to the system concerned.
General information, not legal advice.
Report an errorReferences (4)
- Regulation (EU) 2024/1689 of 13 June 2024 laying down harmonised rules on artificial intelligence (AI Act). (opens in a new tab)
- Regulation (EU) 2026/1744 amending Regulation (EU) 2024/1689 ("Digital Omnibus on AI"), Official Journal of the EU of 24 July 2026, in force since 27 July 2026. (opens in a new tab)
- Law No. 09-08 on the protection of individuals with regard to the processing of personal data (Dahir No. 1-09-15 of 18 February 2009), in particular Articles 4, 11, 43 and 44. (opens in a new tab)
- ISO/IEC 42001:2023 — Information technology — Artificial intelligence — Management system. (opens in a new tab)
