Hannah Arendt’s account of the "banality of evil,” developed while she was observing the trial in Jerusalem of the Nazi officer Adolf Eichmann, became one of the most widely debated ideas in modern political and moral thought. What primarily caught Arendt’s attention was that she did not encounter in Eichmann – one of the key actors in a machinery responsible for the deaths of millions – the kind of demonic or extraordinary personality she had expected. Instead, what she saw before her was an ordinary bureaucrat who regarded himself as simply following orders and doing his job. This was precisely what Arendt found most disturbing. Great evils did not necessarily have to arise from great hatred or exceptionally wicked individuals. People could become part of immense evil simply by failing to think, by refusing to confront the moral consequences of their actions, and by transferring responsibility to the system within which they operated.

Arendt’s use of the term "banality” certainly did not mean that evil itself had become insignificant. On the contrary, it pointed to the possibility that the source of evil could become ordinary. A person could create a bureaucratic, legal, or psychological distance between himself and what he was doing and then say, "I did not make the decision,” "I was only following orders,” "Everyone was doing it,” or "The rules required it.” In this way, despite participating in the act itself, he could refuse to regard its moral responsibility as his own.

When the concepts developed within Islamic moral thought to explain how the human relationship with evil can change are considered together, it becomes possible to situate the mechanism identified by Arendt within a broader moral framework. Beginning with "ghaflah" ("heedlessness"), reinforced through repetition and "ulfah" (familiarity), and continuing through the weakening of moral sensitivity and "tazyin" ("the embellishment or beautification of wrongdoing"), this process may ultimately reach a point that can be described as the "normalization of 'munkar,' as "'ma‘ruf.'"

What is meant here, of course, is not that "munkar" morally becomes "ma‘ruf" – that evil turns into good. Evil remains evil, but the human relationship with it changes. A behavior that initially shocks, angers or moves a person to action gradually becomes familiar, ordinary and eventually a normal part of everyday life.

A person, of course, does not wake up one morning having suddenly lost all moral sensitivity. First comes "ghaflah," then repeated exposure leads to habituation to evil, followed by silence, and eventually by the production of justifications for what one does or merely witnesses. Evil is thus legitimized in the name of security, order, necessity, self-interest, duty, progress or some greater good. Ultimately, although the objective gravity of evil remains unchanged, the moral response to it progressively weakens. For a society, therefore, the real danger lies not merely in the proliferation of evil, but in people losing their capacity to be shocked by it and to regard it as abnormal. The process by which "munkar" comes to be treated as "ma‘ruf" is completed not when evil becomes more widespread, but when evil ceases to disturb us.

At this point, Islamic moral thought raises another important question that takes the issue of responsibility beyond Arendt’s analysis: Are human beings responsible only for the evil they themselves commit, or are they also responsible for their response to the normalization of evil? It is precisely here that the principle of "al-amr bi'l-ma'ruf wa al-nahy 'an al-munkar," enjoining what is right and forbidding what is wrong, becomes relevant. In the Prophet Muhammad’s well-known hadith on this matter, believers are instructed to change evil with their hand if they are able; if that is not possible, then with their tongue; and if even that is beyond their capacity, then to reject it in their heart.

The capacity to intervene may diminish, but the moral distance between the individual and evil must never be allowed to disappear. Even when a person is unable to prevent evil, he must preserve within himself the conviction that it is wrong. Thus, a person’s capacity for intervention may fall all the way to zero, but his moral position must never be allowed to fall to zero, for the heart is the final stronghold.

It is precisely at this point that the digital age adds a new and far more powerful dimension to the banalization of evil. Today, we follow a significant portion of what is happening around the world through the constantly refreshing feeds of social media platforms, and algorithms largely determine which news stories, images, or comments we encounter.

Consider just a few minutes of scrolling through social media. First, we see images of children killed in a war. We swipe, and a few seconds later we watch our favorite team score a goal. Then comes an advertisement, a bombed-out city, a funny video, people on the brink of starvation, a holiday advertisement, and a piece of celebrity news. All these events, utterly different from one another in moral terms, appear on the same-sized screen, are consumed within seconds, and are left behind with the same movement of the finger.

For the algorithmic feed to banalize evil, it does not even need to tell us, "This is not evil.” It is enough for it to turn evil into an ordinary component of an endless stream of content. While traditional propaganda seeks to change the meaning of evil, the algorithmic feed changes its moral weight. The death of a child and a football goal, images of war and an advertisement, news of starvation and an entertainment video are all subjected, technologically, to the same treatment.

Added to this are speed and repetition. Yet, moral judgment requires time. For a person to pause before another’s suffering, reflect on the causes of what he has witnessed, empathize with the victim, and ask, "Do I have a responsibility here?” requires slowing down. The logic of the algorithmic feed, however, is not to stop but to keep moving. Before the emotions evoked by a tragedy we saw only seconds earlier have had time to settle in our minds, a new image is already layered over them. Moreover, when we encounter the same or similar images hundreds of times, the shock of the first encounter may gradually diminish. Seeing more does not always mean understanding more or becoming more sensitive.

What further amplifies the influence of algorithmic systems is personalized repetition. As people interact with certain types of content, the system shows them more of the same. As a result, individuals not only become accustomed to particular forms of behavior but may also come to believe that such behavior is far more widespread in society than it actually is.

The statement "everyone does it anyway” may thus cease to be merely an excuse people use to reassure themselves and instead become the product of an algorithmically constructed perception of reality. Although likes, views, and shares are not moral criteria, they can function as indicators of social acceptance. In this way, algorithms indirectly influence not only what we see but also what we come to regard as normal.

Yet, with the rapid development of artificial intelligence, this transformation does not end here. AI is no longer merely a tool that provides information to humans; it is increasingly evolving toward agentic systems capable of making decisions on behalf of humans, carrying out tasks, and developing their own intermediate strategies to achieve assigned goals. This raises a new moral question: When people delegate a task to artificial intelligence, do they also delegate the moral responsibility associated with that task?

A recent study by Nils Köbis and colleagues, published in Nature, provides a striking answer to this question. In the experiments, levels of honesty remained high when participants reported outcomes themselves, but declined substantially when the task was delegated to a machine. The most striking result emerged when participants were given the option of selecting only a general objective for the machine. When asked to choose between "maximize accuracy” and "maximize profit,” selecting objectives that prioritized profit led to extremely low levels of honest behavior. This is because the individual is now able to retreat into a psychological space in which he can say, "I did not tell it to cheat; I only asked it to maximize profit.” On the other hand, a person can respond to an unethical order by saying, "I will not do this,” even at personal cost. A machine, however, possesses neither such moral agency nor a conscience. The machine has no "heart.” Its behavior is determined not by moral judgment but by the goals assigned to it, its objective function, its training process, and the safety constraints built into the system. The intermediate moral layer that exists in a human chain of command may therefore disappear when a task is delegated to a machine.

In the age of artificial intelligence, "tazyin" also acquires a new language: optimization. A person can say "I optimized revenue” instead of "I set misleading prices,” "I maximized engagement” instead of "I manipulated people,” or "I increased efficiency” instead of "I implemented an unfair practice.” Thus, while bureaucratic language conceals the moral consequences of an action behind the word "duty,” algorithmic language hides those same consequences behind the word "optimization.” The technical vocabulary may change, but the moral distance that individuals place between themselves and their actions remains.

Within this framework, the banalization of evil in the algorithmic age occurs on two distinct levels. At the first level, the algorithmic feed transforms the individual into a spectator. As war, death, violence, gambling, insults or violations of privacy become continuously visible, people grow accustomed to them and their moral response weakens. At the second level, although delegating tasks to AI may appear to return the individual to the position of an agent, leaving the execution of the act to a machine can make that individual feel more distant from responsibility. In short, while the algorithmic feed risks weakening the spectator’s moral sensitivity, delegation to artificial intelligence risks weakening the agent’s sense of responsibility.

As we can see, technology is fundamentally changing the relationship between human beings and their own moral responsibility. Arendt’s bureaucratic individual said, "I was only following orders.” The individual shaped by the algorithmic feed can say, "I only saw it,” while the individual of the AI age can say, "I only set the objective; the AI decided how to achieve it.” In all three cases, only the intermediary changes; what remains constant is the human tendency to create distance between oneself and the moral meaning of an act and, ultimately, the responsibility it entails. As this distance grows, the conditions under which "munkar" can come to be treated as "ma‘ruf" become stronger. In this way, a new mechanism emerges that makes possible the normalization of "munkar" as "ma‘ruf" on a global scale.

It is precisely for this reason that the Islamic principle of "rejecting evil in one’s heart” acquires a critically important new meaning in the age of algorithms. No matter how advanced technology becomes, human beings must preserve their capacity to recognize evil as evil. Seeing is not the same as looking; being informed is not the same as understanding; watching is not the same as bearing witness; and delegating a task is not the same as delegating responsibility. Human beings may delegate tasks to AI, automate decision-making processes, and leave the details of execution to machines. But they can never delegate their moral responsibility.