As of January 2026, the stocks of seven of the world’s most well-known technology companies had reached market capitalizations of about $1 trillion or more. Dubbed the ‘Magnificent Seven,’ their public image hardly lives up to the name. In fact, quite the opposite is true. Companies like Tesla and Meta are known for causing harm to the public good in many ways. In her discussion of the AI polycrisis, Keiko Tanaka draws on the decades-old concept of kōgai (public nuisance) developed in Japan and reflects on how it helps us conceptualize our current situation.
*
Having spent over twenty years thinking about how digital media technologies can serve the public interest, it’s time to revisit the concept of kōgai, which was developed by activist/scholar-practitioner Jun Ui half a century ago.
The term kōgai was originally coined to describe public nuisance. Today, it aligns remarkably well with today’s troubling AI polycrisis. Issues such as job losses due to automation, eroded mental health from algorithmic feeds, undermined democracies, deepfakes, soaring electricity bills, and the climate impact of AI data centers exemplify what I refer to as digital kōgai.
What is kōgai?
The term kōgai originally meant the opposite of ‘public benefit’ or ‘public interest,’ indicating something that damages the common good. This idea developed as part of a movement that fought against the injustices of that era, particularly those arising from Japan’s rapid modernization and industrialization in the 1950s and 1960s. Environmental destruction became a prominent feature during this period and was later recognized as a cause of disease. It took roughly a decade for legislative measures and relief to be put in place.
The following key features of the kōgai remain relevant today. First, the term indicates harm to large groups of people and communities. Second, the sources of harm are often numerous and difficult to pinpoint. Third, even when a source can be identified, establishing a direct causal link between that source and the harm is extremely challenging. This makes it nearly impossible to obtain judicial relief.
These characteristics were documented in a 1966 report investigating neurological paralysis, excruciating bone pain, and respiratory illness endemic to communities near industrial facilities. Recognizing the difficulties victims faced in seeking relief, the report paved the way for the development of new legal frameworks.
The situation sounds strikingly similar to some symptoms of digital harm. Most AI polycrises are endemic to certain demographics (e.g., body image issues for teenagers due to social media or job loss for those in highly routinized manual or cognitive work due to AI automation), matching the first feature. The difficulty of identifying the sources of digital harm has long been debated (e.g., the Section 230 argument, the concept of ‘many hands,’ ‘datafication,’ or the question of whether the responsibility lies with AI companies or businesses demanding the expansion of AI model advancements). Few court cases have successfully established causation, matching the third feature.
Jun Ui further articulates the concept of kōgai as the result of prioritizing profits over community autonomy. In turn, Shoji and Miyamoto described kōgai as having the following characteristics: the anticipation of large-scale harm due to the generation and accumulation of pollutants; the prioritization of profit over safety measures; the promotion of a consumer lifestyle; and a legislative void in prevention. As a result, kōgai impairs the daily lives of the population.
Does anyone else think that a mother’s distress over her toddler’s screen addiction perfectly exemplifies the features of kōgai, given that the legislature, phone companies, and content providers care little about children’s development?
Kōgai paradigm vs. pollution paradigm
Although kōgai is sometimes translated as ‘environmental pollution,’ it is paradigmatically different from the concept of pollution. This distinction is important because the term is also associated with misinformation and disinformation, which could be considered “digital pollution.” However, I believe this is an incorrect diagnosis of the problem.
An article uses the term “digital pollution” to highlight the danger of neglecting online hate speech and trolls. The article draws an analogy between the cholera outbreak in 19th-century London and today’s ungoverned misinformation on the internet, referring to the latter as “digital pollution.” Referring to the pollution during the Industrial Revolution and the construction of the sewage system as the solution, the article writes, “Society figured out how to manage the waste produced by the Industrial Revolution. We must do the same thing with the internet today.”
However, I find the pollution paradigm problematic. It frames harm as a natural byproduct of progress – neutral, inevitable, and solvable through technical innovation. This narrative relies on the familiar trope of the white male savior engineer while erasing the local socioeconomic complexities and power dynamics at play. In doing so, it fails to effectively diagnose the problem, reducing a multifaceted socio-structural issue to a mere technological fix.
Jun Ui’s Kogai paradigm precisely criticizes this and calls for community involvement in the process, similar to today’s desgin justice principle.
In 1992, Jun Ui wrote: “Problems caused by a ravaged human environment are not solvable through the mere application of a technical ‘fix’ or the introduction of new legal structures that ignore fundamental local differences and the need for the participation of pollution victims in the problem-solving process. […] The most important factor for the prevention of pollution problems is the development in the general population of an appreciation for basic human rights and the need always to remain free from oppression.”
Today, we know that those cleaning up digital pollution are not the engineers who design the platforms, but rather, the low-wage workers who scrub them. The 2018 documentary film “The Cleaners” exposed the lives of content moderators in the Philippines who review up to 25,000 posts a day and endure graphic violence and abuse. This has caused them trauma, leading to depression and suicide.
Jun Ui theorized that for every kōgai, there is discrimination. This phenomenon is no exception: the well-being of these workers is not prioritized; rather, the focus is on the uninterrupted flow of content that generates profit, and the burden is disproportionately borne by those with the least power.
AI can fix it?
You might argue that these problems will be solved as AI models become more sophisticated. However, if we look at the full scope of the AI development supply chain, we see that it operates in the same manner.
In her lecture on AI hype, ethics, and algorithmic racial bias, Timnit Gebru discusses the injustices uncovered by investigative journalists Andrea Paola Hernández and Karen Hao, who found that data annotation workers in Kenya are paid poverty wages to label toxic content for AI companies. These workers struggle to feed their children and face eviction. The same logic applies from the poisoned farmers of the Ashio Copper Mine to today’s content moderators, data labelers, and communities bearing the cost of data centers. This is not a flaw in the system – it is how the system operates. We need to consider the entire life cycle of the AI supply chain (as suggested by Kate Crawford), from extraction to training to disposal. We must recognize that kōgai operates at every stage, invisible to those who benefit and devastating to those who bear the cost.
Jun Ui offers another powerful principle: “There are no third parties to kōgai.”
According to this principle, no one can claim true neutrality when it comes to causing harm. He sharply criticized experts, government officials, and other professionals who presented themselves as neutral observers, arguing that they almost always ended up serving as mouthpieces for polluters. His point extended beyond privileged authority. He argued that the next pollution incident would force ordinary citizens to take a side, either as victims or as perpetrators. According to him, the entire population is made up of potential parties to the issue because industrial society entangles everyone in its web of production, consumption, and waste.
Mindful of Ui’s warning that passivity itself is a form of complicity, many of us are trying our best to take action – to take a stand where we can, however limited our reach may be.