Skip to content
Data Protection

Automated decision-making & Data Protection

By Haymanot & Advocates 7 November 2025 5 min read

Introduction

In the digital era, decisions traditionally made by humans are increasingly delegated to algorithms, AI, and data-driven systems. Two core concepts in this transformation are automated decision-making, where outcomes are determined with minimal or no human involvement, and profiling, the processing of personal data to evaluate specific aspects of an individual. While profiling may support decisions, it does not always lead to automated decision making, and vice versa. A key concern in this context is the opacity of such systems, which can obscure how decisions are made and limit opportunities for data subjects to contest them.

Ethiopia’s Personal Data Protection Proclamation 1321/2021 (the “PDPP”) addresses these challenges by regulating automated decision-making and putting safeguards in place to ensure such processes are transparent, fair, and subject to human accountability. This article concisely examines the PDPP framework and outlines essential measures for lawful and responsible employment of automated systems for decision-making.

Understanding automated decision-making and profiling.

Automated decision-making refers to processes where technology makes decisions entirely without human involvement. Under the PDPP, an automated device is defined as a device that makes decisions, initiates, or implements actions based on specific environmental conditions, time, or data, using a pre-arranged algorithm and configuration, without human intervention.

Automated decision-making and profiling, while distinct, often intersect. Profiling involves the automated processing of personal data to assess or predict various personal aspects, such as an individual’s economic situation, health, preferences, or behavior. When incorporated into automated decision-making, profiling enhances the decision-making process by enabling dynamic, data-driven analysis. For instance, a bank’s automated loan approval system might initially consider basic criteria like income but, through profiling, it can integrate additional data such as borrowing behavior or payment history, leading to more informed, yet fully automated, decisions.

While both profiling and automated decision-making analyze data, they differ in focus and impact. Profiling aims to categorize or predict behaviors based on personal data, whereas automated decision-making directly makes decisions based on that data, without human oversight. For instance, a bank using an algorithm to approve a loan based solely on credit score and income is automating the decision. In contrast, if the bank incorporates a broader review of financial history with human judgment, the process involves both profiling and human input, not full automation.

The PDPP regulates both Automated Individual Decision-Making and Profiling. The use of the comma separating “automated processing” and “including profiling” suggests that profiling is one of two distinct criteria for application, with the other being automated processing, which may or may not involve profiling. Specifically, the PDPP addresses the rights of data subjects regarding decisions made solely through automated processing, including profiling, that significantly affect the individual. It also provides data subjects with the right to request human intervention from the data controller in such cases.

General Rule Against Automated Decisions Making

Data subjects have the right not to be subject to decisions based solely on automated processing, including profiling, that produce legal effects or significantly affect them. They also have the right to request human intervention from the data controller and to express their views on such decisions. 

The right applies when three conditions are met: (i) a decision is made, (ii) based solely on automated processing or profiling, and (iii) leads to legal or similarly significant effects on the data subject.

Solely Automated Processing

The restriction on automated decision-making applies only where decisions are made solely through automated processing, i.e., without any real or substantive human involvement. While the Proclamation does not define “human intervention,” it is clear that mere formal approval of an automated outcome is insufficient. Meaningful intervention must involve a qualified individual with decision-making authority who can critically assess the decision’s fairness and context, and is empowered to alter or override the result where appropriate.

The GDPR, which has influenced the PDPP, offers guidance on human involvement in automated decision-making. As clarified in its working documents, genuine human involvement requires more than mere formal oversight; it demands “meaningful” human intervention by an individual with the authority and competence to modify the decision, and who takes into account all relevant data. Therefore, drawing from this, an automated process does not fall under the restrictions of Article 31, if it functions solely as a decision-support tool, with the final decision being made by a human who reviews the results of the automated process rather than being guided solely by the automated output. This remains true even if the majority of the preliminary decision-making process is automated.

Decisions With Significant/Legal Effect

In addition, for the automated decision-making safeguards under the PDPP to apply, the decision must produce a legal effect or significantly affect the data subject. A legal effect arises when the decision alters an individual’s legal rights or obligations; such as terminating a contract, revoking a license, or denying access to a service. Similarly, decisions that materially impact a person’s life, such as those affecting employment, credit, housing, or public benefits, fall within the scope. Conversely, decisions with only minor or inconsequential consequences do not trigger the protections under this provision.

Exception

The right not to be subject to automated decision-making under the PDPP is not absolute and may be waived in three specific cases: (a) where the decision is necessary for entering into or performing a contract with the data subject; (b) where it is authorized by law, provided the law includes appropriate safeguards to protect the data subject’s rights and interests; or (c) where the data subject has given explicit consent.

Rules for Processing Under The Exceptions

Where a decision based solely on automated processing falls within the permitted exceptions under Article 31(2), the controller must still observe heightened safeguards. Specifically, such decisions must never be based on sensitive personal data. In all cases falling under the exception, the controller is also required to inform the individual of the existence of such processing and its potential impact, ensuring transparency at the outset. Where decisions are based on consent or contractual necessity, the controller must implement appropriate safeguards to protect the individual’s rights, freedoms, and legitimate interests, including measures that allow for human oversight and meaningful recourse.

Penalties

A data controller/processor who fails to honor the data subject’s right not to be subject to automated decision-making—alongside failures to erase data, respect objections to processing, or restrict processing—is subject to criminal liability. The penalty ranges from three to five years’ imprisonment, a fine between 100,000 and 200,000 birr, or both, reflecting the serious legal weight attached to violations of core data protection rights.

Conclusion:

As automation continues to reshape decision-making processes across sectors, it is critical to balance technological advancements with robust safeguards for individuals’ rights. The PDPP establishes clear guidelines for automated decision-making, ensuring transparency, fairness, and accountability. While exceptions exist for contract performance, legal authorizations, and explicit consent, data controllers adhere to heightened safeguards, particularly when sensitive data is involved. The provisions outlined in the PDPP are designed to protect individuals from undue impacts of automated decisions while holding data controllers accountable for compliance. Failure to uphold these rights, particularly the prohibition on unjust automated decisions, carries significant penalties, underscoring the importance of responsible data processing practices in today’s digital landscape.

This article is general information and not legal advice. Contact us about your specific situation.

H&A

Published by

Haymanot & Advocates

Legal update · Data Protection practice