AI Library
The Double Structure of Digital Sovereignty
Europe’s Departure from Palantir and the Chains of American Big Tech
Kim Kyung-jin, Attorney at Law
This is a record of 2026, when European intelligence agencies and defense ministries began removing analytics tools from America’s Palantir. It covers the replacement decisions made by France’s General Directorate for Internal Security (DGSI), Germany’s Federal Office for the Protection of the Constitution (BfV), and the Netherlands Ministry of Defense; the incident in which US export controls severed an ally’s ac…
New English Edition
Artificial Intelligence in Horticulture
Kim Kyung-jin, Attorney at Law
Across five chapters and ten sections, this book examines computer vision for crop diagnosis, harvesting robots and autonomous field systems, smart greenhouses and digital twins, precision irrigation and supply-chain quality control, high-throughput phenotyping, and predictive breeding.
New English Edition
Artificial Intelligence in Food Crop Agriculture
Kim Kyung-jin, Attorney at Law
Across six chapters and eighteen sections, the book examines digital agricultural infrastructure, remote sensing, crop diagnosis, yield forecasting, precision irrigation, genomics, molecular breeding, agricultural robotics, climate-smart agriculture, and global food security.
New English Edition
The Future of Forestry and Agroforestry
Kim Kyung-jin, Attorney at Law
Driven by Artificial Intelligence and Digital Innovation
Across five chapters and fifteen sections, the book follows satellites, drones, LiDAR, digital twins, forest-specific language models, wildfire and pest forecasting, forestry robotics, agroforestry, timber traceability, and forest carbon markets.
New English Edition
Smart Livestock Farming: AI Enters the Barn
Kim Kyung-jin, Attorney at Law
Sensors listen, cameras watch, and artificial intelligence helps farmers decide.
Across five chapters and fifteen sections, the book follows precision livestock farming from animal health and reproduction to robotic milking, virtual fencing, digital twins, methane reduction, welfare, and data ownership.
Table of Contents
Han Dong-hoon, Busan Buk-gu Gap: A Record of the 100 Days Before and After the Election (Mar. 26-Jul. 3, 2026)
Kim Kyung-jin
Table of Contents and 13 sections
From March 26 to July 3, 2026, this record follows the spring after expulsion, the Busan Buk-gu Gap by-election, victory as an independent, and the first bill submitted in the National Assembly.
Table of Contents
Artificial Intelligence and Medicine
Kim Kyung-jin, Attorney at Law
AI in clinical care, hospitals, education, and research
AI in medical imaging, risk prediction, treatment planning, hospital operations, education, and research, with patient safety, privacy, and accountability.
[AI Library] Chapter 8. Artificial Intelligence and Data
Ten Questions AI Poses to Humanity
Chapter 8. Artificial Intelligence and Data
Kim Kyung-jin
Garbage in, garbage out
Behind the polished, intelligent responses of ChatGPT, Claude, and Gemini that we talk to every day lies a dark truth most of us never see. Behind every smooth, seemingly flawless sentence an AI produces, there is labor that eats away at someone's mental health, a biased gaze baked into the system, and eyes quietly stealing glances at our private lives. This is the unvarnished face of the digital age we live in.
1. The Suffering of AI Data Workers in Kenya
In an office in Nairobi, Kenya, 27-year-old Mophat Okinyi reviews 700 texts a day. They contain horrific descriptions of rape, murder, torture, and child sexual abuse. He has to read material like this all day, every day. Why? To train ChatGPT to filter out such content.
'It completely destroyed my mental health,' Okinyi says. After reading text after text about rapists, he began avoiding people around him and started viewing everyone with paranoid suspicion. His pregnant wife eventually told him, 'You've changed,' and left. 'I lost my family,' he said, words that reveal the real cost of the AI we use.
Okinyi's story is not unique. According to a 2024 investigation by The Guardian, thousands of 'content moderators' in Kenya alone review this kind of horrific material every day for less than two dollars an hour. They watch beheading videos, classify child sexual abuse images, and label suicide footage. All so the AI we use can be 'safe.'
Faith (a pseudonym), a data worker in Kenya, testified: 'I started because the pay was better, but I soon realized I had to simulate horrific conversations. I even had to describe acts of cannibalism in hypothetical scenarios.' Another worker, Stacy, said with frustration: 'The mental health support the company provides is a 30-minute counseling session once a month. They tell you to watch funny videos on TikTok to erase the bad memories.'
In December 2024, CBS's 60 Minutes program exposed their reality, reporting that Kenyan workers were 'overworked, underpaid and exploited, with inadequate mental health support.' At a town hall meeting of the African Content Moderators Union held in Nairobi in November 2024, participants discussed 'burnout and health problems caused by mental and physical distress, long working hours, exposure to graphic content, and lack of ergonomic support.' One participant remarked bitterly, 'One of the perks of this job is that it's never boring.'
Kenyan human rights activist Odanga Madung strongly criticized the situation, calling it 'modern-day slavery.' On May 1, 2023, International Workers' Day, about 150 current and former content moderators gathered in Nairobi and formed the African Content Moderators Union (ACMU). They submitted a petition to the government demanding an investigation into exploitative working conditions.
What is even more shocking is that this is happening all over the world. The same kind of work takes place in East Africa, India, the Philippines, and even in the Dadaab refugee camp in Kenya and the Shatila camp in Lebanon. Educated, multilingual refugees are doing this work for extremely low wages.
Behind the ChatGPT we say 'thank you' to are people like Okinyi, who says: 'I think of myself as a soldier. A soldier takes bullets for people.' We use 'clean' AI built on top of their sacrifice.
2. Biased Data Creates an Unfair World
In 2024, researchers at the University of Washington conducted a striking experiment. They gave three state-of-the-art AI models identical resumes, changing only the names. The results were grim. The AI favored white-sounding names 85% of the time and female names only 11% of the time. A Black male name was never preferred over a white male name. Not once.
This is not just a laboratory finding. It is happening in the real world. In May 2024, the U.S. District Court for the Northern District of California certified a class-action lawsuit against an AI hiring tool. In Mobley v. Workday, the court ruled that 'if AI software participated in hiring decisions, its bias can serve as grounds for a discrimination lawsuit.'
Amazon abandoned its AI recruiting system in 2018 after discovering that the system was systematically excluding women from technical job applicants. Trained on historical hiring data dominated by men, the system had learned to favor words men typically used, such as 'executed' and 'captured.'
Bias in facial recognition technology is even more severe. It identifies white male faces with 99% accuracy but achieves only 65% accuracy for Black female faces. These errors cause innocent people to be flagged as criminals at airport security checkpoints and lead police investigations toward the wrong suspects.
Bias in healthcare is a matter of life and death. A major U.S. healthcare algorithm assigned lower risk scores to Black patients than to white patients, even when their health conditions were identical. The algorithm used healthcare spending as a proxy for health status, ignoring the fact that structural inequality means Black people often have less access to medical services.
Heart disease diagnostic AI misses symptoms in female patients, and skin cancer diagnostic AI fails to properly detect lesions on patients with darker skin tones. Most medical data has been collected primarily from white male patients.
Language processing AI shows the same pattern. Google Translate frequently renders 'doctor' as male and 'nurse' as female. When translating the Turkish sentence 'O bir doktor' (he/she is a doctor) into English, it becomes 'He is a doctor.' Translate 'O bir hemşire' (he/she is a nurse), and you get 'She is a nurse.'
A 2024 report by the European Union Agency for Fundamental Rights (FRA) warned that 'audio algorithms contain strong biases based on ethnicity, gender, religion, and sexual orientation.' The report, published in May 2025, urged 'EU legislators and member states to ensure consistent and high levels of anti-discrimination protection across all grounds.'
Decisions made by biased AI become new data that trains the next generation of AI to be even more biased, creating a vicious cycle. Recommendation algorithms on online platforms trap users inside information bubbles, deepening social division. News recommendation systems show users only content that matches their existing views, cutting off opportunities to encounter different perspectives.
According to a 2024 report by UN Women, 'AI systems learn from data saturated with prejudice, reflecting and reinforcing gender bias. Such bias can limit opportunity and diversity in decision-making, hiring, loan approvals, and legal judgments.'
The problem is that reducing bias requires more money and time, but companies prioritize speed to market and profit. Government regulation cannot keep up with the pace of technological development. Individuals and groups who suffer harm face an uphill battle proving damages and obtaining compensation.
3. Personal Data and Privacy
2024 was the year legal battles over AI and personal data protection exploded. In U.S. federal courts alone, more than 1,970 data privacy lawsuits were filed, a sharp increase from 2023.
In 2023, an AI chatbot introduced at a U.S. hospital was found to have been transmitting patients' sensitive information externally. Symptoms, names, dates of birth, and medical histories that patients had entered were being stored as log files on an external cloud server. Patients had been talking to the AI without any idea where their information was going.
What is even more alarming is that generative AI itself carries the risk of personal data leaks. A 2024 study confirmed cases in which GPT models generated sentences from their training data that included real names, email addresses, and phone numbers. The AI was randomly spitting out personal information it had 'memorized' during training.
In 2024, Washington State enacted the My Health My Data Act (MHMDA), the first comprehensive state law protecting consumer health data beyond the scope of HIPAA. As soon as the law took effect, the first lawsuit was filed against Amazon, alleging that health-related apps and online health services had been collecting health data without user consent.
Self-driving cars have become a breeding ground for new privacy violations. Vehicles from Tesla, BMW, and others collect drivers' locations, driving patterns, and even conversations inside the car. In 2024, the Texas Attorney General filed a lawsuit against General Motors, alleging that GM had sold drivers' driving data without authorization.
What's more alarming is the emergence of neural data. In 2024, California and Colorado amended their privacy laws to include neural data within their scope of protection. VR devices, brainwave monitoring equipment, and wearable devices have begun collecting our brain activity and thought patterns.
In 2024, the Federal Trade Commission (FTC) took successive enforcement actions against data brokers that sold location data. InMarket, X-Mode, Mobilewalla, and Gravy Analytics were all penalized for selling precise location data on sensitive locations without consumer consent.
The problem is that this collection and use of personal information happens without our knowledge. LinkedIn began using user data for AI training in September 2024, but most users had no idea. Those who found out and wanted to opt out had to navigate through a maze of settings menus.
China enacted the world's first AI regulation in 2023. The Interim Measures for the Management of Generative Artificial Intelligence Services require AI companies to strengthen personal data protection. The United States, by contrast, has the Blueprint for an AI Bill of Rights released by the White House in 2022, but it is a nonbinding guideline with no legal force.
The smarter artificial intelligence becomes, the greater the threat to our privacy. Do we really know how the data we hand over for the sake of convenience is being used?
4. The Price of a Single Book for Copyright Fees: Is That Enough?
On June 24, 2025, a federal court in California handed down one of the most significant rulings in the history of artificial intelligence. Judge William Alsup ruled in a copyright lawsuit brought by authors against Anthropic that 'purchasing a single book and using it for AI model training does not constitute copyright infringement.'
The ruling sent shockwaves through the AI industry. Judge Alsup recognized AI training as 'transformative use,' stating that 'similar to how a reader learns to become a writer, AI does not copy the original work but uses it as a foundation for new creation.'
But the ruling came with an important caveat. The judge acknowledged that Anthropic had 'downloaded millions of copyrighted books for free from internet piracy sites' and referred that issue to trial in December 2025.
Then, three months later on September 5, 2025, surprising news broke. Anthropic had reached a $1.5 billion (approximately 2 trillion won) settlement with the authors. Under the terms, Anthropic agreed to pay $3,000 per book for roughly 500,000 titles.
This was the first large-scale settlement over AI and copyright. Legal analyst William Long observed that 'had the trial continued, Anthropic could have faced billions of dollars in damages, possibly enough to bankrupt the company.'
The root of the issue was the source of AI training data. Most large language models, including Anthropic's Claude, were trained on a dataset called Books3. This dataset contained millions of books collected from illegal book sites such as Library Genesis (LibGen) and Pirate Library Mirror.
The scale of data that OpenAI's GPT-3.5 trained on is staggering. 410 billion tokens collected from the web. 19 billion additional web text tokens. 67 billion sentences extracted from books. 3 billion words from Wikipedia. Nearly all of it was used without the original creators' permission.
The Studio Ghibli controversy exposed another dimension of this problem. When OpenAI released ChatGPT-4o's image generation feature in 2024, a global trend of creating 'Ghibli-style' images erupted. Even Sam Altman, OpenAI's CEO, changed his profile picture to a Ghibli-style image.
Behind that craze, though, lay tangled copyright issues. Director Hayao Miyazaki, after seeing AI animation in a 2016 NHK documentary, called the work 'an insult to life itself' and said he had 'absolutely no intention of incorporating this technology into my work.'
Artist Karla Ortiz called it 'yet another clear example that companies like OpenAI do not care about artists' work and livelihoods.' She is pursuing copyright lawsuits against other AI image generation companies.
The news industry is pushing back as well. Jason Conti of News Corp, the parent company of The Wall Street Journal, pointed out that 'if you want to train AI using articles written by Wall Street Journal reporters, you need to obtain proper permission,' adding that 'OpenAI has not signed a licensing agreement with us.'
Similar concerns have been raised in South Korea. The Korea Newspaper Association argued that 'using news content for AI training infringes on the copyright and database producer rights of media companies.'
AI companies claim such practices fall under fair use. But the Korea Newspaper Association countered that the use 'does not perform a new and different function distinguishable from the original work and can substitute for demand for the original,' arguing that 'it does not qualify as transformative use and therefore fair use does not apply.'
Numerous copyright lawsuits were filed in 2024. The New York Times vs. Microsoft and OpenAI, Getty Images vs. Stability AI, Universal Music Group vs. Anthropic, and record labels vs. Suno and Udio, among others. In June 2025, Disney and NBCUniversal sued Midjourney, alleging unauthorized use of characters including Elsa, the Minions, Darth Vader, and Homer Simpson.
Most recently, in September 2025, Warner Bros. Discovery filed a lawsuit against Midjourney in a California federal court. Meanwhile, in August 2025, Elon Musk's xAI sued Apple and OpenAI, showing that legal disputes are spreading even within the AI industry itself.
Still, some rulings have gone in favor of AI companies. In June 2025, Meta won the lawsuit brought by Sarah Silverman and Ta-Nehisi Coates. Judge Vince Chhabria granted Meta's motion for summary judgment.
Against this backdrop, Anthropic's $1.5 billion settlement set an important precedent. It is expected to influence other pending lawsuits. But a fundamental question remains. If AI learns from all of humanity's creative works to produce something new, what compensation should the original creators receive?
We are intoxicated by the convenience and efficiency that artificial intelligence delivers. But behind it all, a young worker in Kenya suffers from mental illness, algorithms discriminate by race and gender, our private lives are collected without consent, and creators' rights are being violated.
If technological progress is to improve human well-being, we cannot look away from these darker realities. For artificial intelligence to become a technology that truly benefits everyone, we need a development culture and social consensus that values fairness, inclusion, and human dignity just as much as technical performance.
Garbage in, garbage out. Feed in bias and exploitation, and bias and exploitation come out. The data we choose to put in will determine the AI of the future, and the world of the future.
Kim Kyung-jin
Attorney · Former Member of the National Assembly · AI Policy Researcher
© 2026 Kim Kyung-jin. All rights reserved.