5.2: AI Ethics, Digital Repression and Global Case Studies

Artificial intelligence does not arrive in society as a neutral tool. It arrives embedded in systems of power, shaped by histories of inequality, and quickly becomes part of the invisible infrastructure that governs everyday life. To understand AI today is not simply to understand a technology, but to understand a transformation in how decisions are made, how authority is exercised, and how human futures are determined.

Consider a seemingly ordinary moment. A recent graduate submits dozens of job applications online. Within seconds, automated systems scan resumes, rank candidates, and filter out applicants deemed “less suitable.” No human ever reviews many of these applications. Somewhere in that process, an algorithm has already made a judgment about potential, capability, and worth. Now multiply this moment across millions of decisions: who gets hired, who receives a loan, who is flagged for further screening at an airport, who is denied insurance, who is predicted to reoffend. Increasingly, these decisions are shaped not by human deliberation but by systems trained on vast datasets, optimized for efficiency, and shielded from scrutiny.

This is the world Cathy O’Neil (2016) warns about in her concept of “Weapons of Math Destruction.” These are systems that are opaque, scalable, and capable of producing harm at massive scale, all while maintaining the appearance of objectivity. The danger lies not only in individual errors, but in the normalization of automated judgment. When decisions are framed as the outcome of data-driven processes, they are often perceived as neutral, even when they replicate deep structural biases.

The core problem is deceptively simple: AI systems learn from data, and data reflects history. But history is not neutral. It is shaped by inequality, discrimination, and uneven access to opportunity. When AI systems are trained on historical data, they do not merely reflect the past. They can reproduce and amplify it, encoding inequality into the very systems that shape the future.

This raises a profound question about the nature of fairness itself. In traditional decision-making, bias can at least be contested. A human decision-maker can be questioned, challenged, or held accountable. But when decisions are made by complex algorithms, accountability becomes diffuse. Responsibility is distributed across developers, companies, and institutions, making it difficult to identify where harm originates or how it can be addressed.

The work of Joy Buolamwini and Timnit Gebru (2018) offers a powerful illustration of this problem. Their research on facial recognition systems revealed that these technologies performed significantly worse when identifying darker-skinned women compared to lighter-skinned men. This was not simply a technical glitch. It was a reflection of whose data had been prioritized in the training process. Faces that were already overrepresented in datasets were recognized more accurately, while others were systematically misidentified.

But the implications go beyond accuracy. Misidentification in facial recognition systems can lead to wrongful arrests, denial of services, or exclusion from opportunities. What appears as a technical issue becomes a question of justice. Who is seen clearly by the system, and who is rendered invisible or misrepresented?

As governments and corporations deploy AI systems at increasing scale, the concept of algorithmic accountability becomes central. Yet accountability in this context is not straightforward. Many advanced AI models operate as black boxes, meaning that even their creators cannot fully explain how they arrive at specific decisions. This creates a tension between performance and transparency. The systems that perform best are often the least interpretable.

Efforts to address this challenge are emerging across different contexts. The European Union’s AI Act introduces a risk-based framework, categorizing AI systems based on their potential impact and requiring stricter oversight for high-risk applications. UNESCO’s Recommendation on the Ethics of AI, adopted by 193 member states, emphasizes principles such as fairness, transparency, and human oversight. These initiatives signal a growing recognition that AI governance must extend beyond technical design to include ethical and social considerations.

Yet regulation alone cannot resolve these challenges. At its core, algorithmic accountability is a question of power. Who decides what counts as fair? Who has access to the data and systems that shape decision-making? And whose voices are included in the design and governance of these technologies?

These questions become even more urgent when we consider the impact of AI on labor. Automation has long been associated with technological progress, but the current wave of AI-driven transformation is unprecedented in its scope and speed. Tasks that once required human judgment are increasingly automated, affecting not only manual labor but also cognitive work.

The World Economic Forum (2023) estimates that 44 percent of workers’ skills will be disrupted within the next five years. This does not mean that jobs will simply disappear. Rather, the nature of work itself is being reconfigured. Some roles are eliminated, others are transformed, and new ones emerge. But these changes are uneven. Workers in certain sectors and regions are more vulnerable to displacement, while others benefit from new opportunities.

At the same time, AI is creating new forms of labor that are often hidden from view. Behind every AI system are workers who label data, moderate content, and maintain the infrastructure that makes automation possible. These workers are frequently located in the Global South, where labor costs are lower and regulatory protections may be weaker. They perform essential tasks, yet their work remains largely invisible within dominant narratives of technological innovation.

This creates a paradox. AI is often portrayed as replacing human labor, but in reality it relies on vast amounts of human effort. The difference is that this labor is distributed, precarious, and often undervalued. As a result, discussions about the “future of work” must also address the global inequalities embedded within digital economies.

Beyond labor, AI raises fundamental questions about the right to life itself. Autonomous weapons systems represent one of the most controversial applications of AI. These systems are capable of identifying and engaging targets without direct human intervention. For proponents, they offer strategic advantages and the potential to reduce risks for military personnel. For critics, they represent a dangerous step toward delegating life-and-death decisions to machines.

The ethical concerns are profound. Warfare has always involved difficult moral decisions, but those decisions have traditionally been made by humans who can be held accountable. When an autonomous system makes a lethal error, accountability becomes ambiguous. Is the responsibility with the programmer, the operator, the military command, or the state?

The Campaign to Stop Killer Robots argues that meaningful human control must be maintained over decisions involving lethal force. Yet international consensus on this issue remains elusive. Major powers continue to invest in autonomous systems, viewing them as essential to maintaining strategic advantage.

These debates highlight a broader challenge: as AI systems become more capable, the boundaries of human responsibility become increasingly blurred. The question is not only what machines can do, but what humans are willing to allow them to do.

While AI raises ethical dilemmas in democratic contexts, it also enables new forms of authoritarian governance. Digital technologies provide states with powerful tools for surveillance, censorship, and social control. In some contexts, these tools are used to consolidate power and suppress dissent.

China offers one of the most comprehensive examples of digital authoritarianism. Its approach combines extensive surveillance infrastructure with advanced data analytics and AI-driven governance systems. The Great Firewall restricts access to foreign information, while domestic platforms operate under strict regulatory oversight. The Social Credit System integrates data from various domains to evaluate and influence behavior.

What is striking about this model is not only its scale, but its integration into everyday life. Surveillance is not limited to exceptional circumstances. It is embedded in routine interactions, from financial transactions to social media activity. This creates a system in which behavior is continuously monitored and shaped through incentives and penalties.

At the same time, China’s model is not simply about repression. It is also framed as a form of governance, aimed at enhancing efficiency, stability, and trust. This dual framing complicates simplistic narratives about authoritarianism. It suggests that digital control can be justified and even normalized within certain contexts.

India presents a different but equally complex case. As the world’s largest democracy, it hosts a vibrant and diverse digital public sphere. Yet it has also implemented extensive digital infrastructure, including the Aadhaar biometric identification system, which covers over a billion people.

Aadhaar has been praised for improving access to services, particularly for marginalized populations. At the same time, it raises concerns about surveillance, data security, and the potential for misuse. The increasing use of internet shutdowns and expansive surveillance powers highlights the tension between democratic ideals and state authority.

Russia’s approach to digital governance reflects a more overt emphasis on control. The Sovereign Internet Law enables the state to manage internet traffic within its borders, allowing for greater censorship and monitoring. During periods of political tension, this infrastructure has been used to restrict access to information and suppress dissent.

Yet in each of these contexts, resistance persists. Activists and ordinary citizens develop strategies to navigate and challenge digital control. They use encrypted messaging, virtual private networks, and creative forms of expression to maintain spaces of autonomy. This ongoing interaction between control and resistance underscores the dynamic nature of digital power.

In the Middle East and North Africa, digital technologies have played a central role in both mobilization and repression. During the Arab Spring, social media platforms enabled rapid coordination and global visibility for protest movements. In the years that followed, many governments adapted, using the same technologies for surveillance and control.

Despite these challenges, digital spaces remain critical for activism. Movements such as those in Iran and Lebanon demonstrate how online networks can sustain resistance even under restrictive conditions. These examples highlight the dual nature of technology as both a tool of empowerment and a mechanism of control.

This duality is perhaps most visible in the rise of cyber-activism. Digital platforms have transformed how movements emerge, organize, and communicate. Hashtags can mobilize global attention within hours. Images and videos can document injustice and circulate across borders.

Movements such as the Arab Spring, the Hong Kong protests, and Black Lives Matter illustrate the power of networked activism. They demonstrate how digital tools can amplify voices, build solidarity, and challenge dominant narratives.

At the same time, these movements reveal the limitations of digital mobilization. As Zeynep Tufekci (2017) argues, the speed and scale of online organizing can come at the expense of deeper organizational structures. Movements may mobilize quickly but struggle to sustain momentum or achieve long-term change.

The result is a complex landscape in which technology both enables and constrains collective action. It lowers barriers to participation while introducing new vulnerabilities, from surveillance to misinformation.

Across regions, different approaches to digital governance reflect broader political and cultural contexts. The European Union emphasizes regulation and rights-based frameworks, positioning itself as a global leader in digital governance. The United States relies more heavily on market mechanisms and corporate self-regulation, leading to ongoing debates about platform responsibility.

In Kenya, digital innovation has expanded financial inclusion through systems such as M-Pesa, while also raising questions about data governance and sovereignty. In Brazil, the spread of misinformation through messaging platforms has challenged democratic processes, prompting new efforts to balance free expression with accountability. In South Korea, advanced technological infrastructure has enabled efficient responses to crises such as the COVID-19 pandemic, while also raising concerns about privacy and data use.

These cases illustrate that there is no single model for managing the relationship between technology and human rights. Instead, there are multiple pathways, each shaped by historical, political, and economic factors.

What connects these diverse experiences is a shared challenge: how to harness the benefits of digital technologies while mitigating their risks. This requires not only technical solutions but also ethical reflection, institutional innovation, and collective action.

At its core, the story of AI and digital repression is a story about power. It is about who designs the systems that shape our lives, who benefits from them, and who is left vulnerable. It is about the tension between efficiency and justice, innovation and accountability, control and freedom.

And it is still unfolding.