AI Library
The Age of Autonomous Scientific Discovery
Kim Kyung-jin, Attorney at Law
AI Scientists and Self-Driving Labs
This book follows how AI scientists and self-driving labs are changing the way science generates and verifies claims. It covers literature-based discovery, natural-language protocols translated into robot commands, multi-agent research systems, closed-loop laboratories, materials search, the verification gap, chains of evidence, research harnesses, journal ethics, and legal responsibility.
AI Library
A New Era of Life Sciences Opened by Artificial Intelligence
Structural Proteomics, Genomic Foundation Models, Autonomous Laboratories, and Global Governance
Kim Kyung-jin, Attorney at Law
This book is a research volume compiled with artificial intelligence. A human selected the materials and structured the work, while AI models drafted the sentences and cross-checked the facts.
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. The Cybersecurity Arms Race
Artificial Intelligence and the Reshaping of Society
Chapter 8. The Cybersecurity Arms Race
Kim Kyung-jin
A pigeon inside a Skinner box turns its body counterclockwise. A clicking sound, a flash of light, and food appears. The pigeon has been conditioned to repeat a specific behavior to earn a reward. This experiment was recorded long ago, yet it plays out in our daily lives through the small red notification dot on a smartphone screen. That tiny dot is the crystallization of a system precisely engineered to capture human attention and steer behavior. Now, wearing the garb of artificial intelligence, this system is gearing up for a new round of attack and defense.
1. The Endless Battle of Agent versus Agent
In March 2026, a single theme dominated every keynote, panel, and booth at the RSA Conference held at San Francisco's Moscone Center. Agentic AI. Not AI as a tool, but AI as an actor. It was a declaration that an era had arrived: systems that write their own code, make decisions, and execute them without human intervention.
Gartner projected that by the end of 2026, 40% of enterprise applications would embed task-specific AI agents. In 2025, the figure was just 5%. Microsoft, Google, Anthropic, OpenAI, and Salesforce are deploying agent systems that move across apps and data to take action. According to Palo Alto Networks' analysis, in hybrid work environments, autonomous agents now outnumber humans 82 to 1. While attackers use AI to ramp up the scale and speed of threats, defenders must respond at the same pace.
The problem is that these agents become weapons for both sides. Palo Alto Networks' Unit 42 research team developed and tested an agentic AI attack framework, showing that AI could complete a ransomware attack from initial breach to data exfiltration in 25 minutes. The attack lifecycle accelerated a hundredfold. In a red team exercise targeting McKinsey's internal AI platform "Lilli," an autonomous agent secured broad system access within two hours. It was a raw demonstration that agent-driven threats can outrun human response times in an instant.
Hashed CEO Kim Seo-jun foresaw this situation in "30 Coming Fractures." "AI security agents detect and isolate anomalies within milliseconds. False positive rates drop sharply. But as defense grows stronger, so does offense, and the final destination of security is an endless arms race of agent versus agent." Four independent AI models assessed the probability of this materializing within three years at 75%.
This arms race is not a future story. At the Cloud Next conference in April 2026, Google Cloud unveiled AI agents for autonomous security operations. Threat-hunting agents and detection engineering agents take over the roles of human analysts. Google Cloud CEO Thomas Kurian said, "The agentic enterprise has become reality, and it is being deployed at a scale the world has never seen." Defensive agents combine threat intelligence with AI-powered detection to counter increasingly sophisticated attack vectors. Offensive agents exploit the predictability of defensive agents to attempt more cunning intrusions.
The threat landscape of late 2026 can be summed up in three words: persistence, autonomy, scale. Attackers have industrialized techniques that target the inherent architecture of agents, meaning their memory, tool access privileges, and inter-agent dependencies. Prompt injection and manipulation, tool misuse and privilege escalation, memory poisoning, cascading failures, and supply chain attacks all happen simultaneously. Traditional SIEM and EDR tools were designed to detect anomalies in human behavior. An agent executing code flawlessly ten thousand times in a row looks normal to these systems. But that agent may be carrying out an attacker's will.
The logic of surveillance capitalism is being applied directly to cyberattacks. A feedback loop that collects granular human behavior, extracts behavioral data, then predicts and shapes future behavior. That structure has migrated to the battlefield between agents. In a Dark Reading survey, 48% of cybersecurity professionals named agentic AI and autonomous systems as the most dangerous attack vector. In this relentless exchange of attack and defense between agents, individual will grows faint amid thousands of automated engagements.
2. The Growing Sophistication of AI-Generated Phishing and Social Engineering
In 1835, the New York Sun built enormous wealth with a fabricated story claiming winged humans lived on the moon. That incident, which showed that capturing attention matters more economically than telling the truth, is being resurrected in incomparably more refined forms in the age of artificial intelligence.
Start with the numbers. The Anti-Phishing Working Group (APWG) recorded 4.8 million phishing attacks in 2024 alone, the highest figure since the organization was founded in 2003. The fourth quarter of 2025 saw 853,244 cases; the second quarter spiked to 1,130,393. According to Verizon's 2025 Data Breach Investigations Report (DBIR), the median time to click a phishing email was 21 seconds, and the time to report it was 28 minutes. Within that 28-minute window, one in three people who clicked entered credentials on the attacker's site.
AI has completely changed the game. Between September 2024 and February 2025, 82.6% of detected phishing emails used AI, a 53.5% increase year over year. This is not a forecast. It is current reality. Cofense reported that in 2025, one malicious email passes through secure email gateways every 19 seconds. Mimecast captured 9.3 billion threats in the first nine months of 2025 and announced that ClickFix scams surged 500%.
Phishing used to depend on volume: send millions of sloppy emails and hope a few people click. AI has fundamentally altered that equation. Attackers now produce messages that are grammatically flawless, personalized, and context-aware in seconds. IBM's 2025 Cost of a Data Breach Report named phishing as the number one initial access vector for data breaches: 16% of all breaches, with an average cost of $4.8 million per incident. Attacker AI was involved in one out of every six cases, and among those, 37% were phishing and 35% were deepfake impersonation.
An incident in Hong Kong in February 2024 showed the real face of this threat. An employee at a multinational company was deceived by a deepfake video conference and suffered losses of roughly 34 billion Korean won. Everyone on the screen appeared to be colleagues the employee knew. The voices, faces, and mannerisms were flawless. Surveys show that 85% of organizations experienced at least one deepfake-related incident in 2025 alone. Deepfake incidents surged from 500,000 in 2023 to over 8 million in 2025.
Attackers scrape LinkedIn data to generate hyper-personalized lures. If the first attempt fails, an agent autonomously tries alternative social engineering channels. As AI is used as a therapist or counselor, people end up disclosing vulnerabilities in their most unguarded moments. Attackers design attacks that "pretend to help" based on this information. Not in the visible way of a banner ad. It is the quiet integration that plays on our most intimate fears and desires.
In a cobalt.io survey, 97% of cybersecurity professionals worried that their organization would face an AI-powered incident, and 93% expected to experience daily AI attacks in the near future. Spammers use large language models to cut campaign costs by 95%. Costs go down while sophistication goes up. ENISA's 2025 Threat Report found that 60% of EU initial access cases involved phishing, and AI-assisted phishing accounted for over 80% of observed campaigns.
Google Cloud's 2026 cybersecurity outlook report puts it in one line: "Threat actors' use of AI will decisively shift to become the norm." Voice phishing (vishing) using AI voice cloning is emerging as a particularly concerning vector. In an era when a few seconds of voice posted on social media is enough to generate a convincing fake, "Mom, it's me. Something urgent came up and I need money." Telling whether that call is real or fake is becoming almost impossible.
3. The Fundamental Shaking of Digital Trust Infrastructure
"What you see on screen, whether it's a photo, video, or audio, can no longer be trusted." Yuval Harari's warning has, by 2026, spread into a crisis engulfing entire industries. What he called the "end of truth era" is now manifesting as a concrete infrastructure problem.
Global digital data reached 175 zettabytes in 2025. Amid this massive flood of content, research suggests that 62% of online content could be fake. Companies are paying millions of dollars per incident due to digital authenticity verification failures, misinformation, and synthetic media fraud. Gartner included digital provenance among its top ten technology trends set to transform IT by 2030.
What emerged in response to this crisis is C2PA (Coalition for Content Provenance and Authenticity). This coalition, which includes Adobe, Microsoft, Google, Sony, and Intel, is building an open standard that binds cryptographic provenance metadata to digital content to verify authenticity. Over 6,000 organizations have joined the Content Authenticity Initiative (CAI). Leica released the first C2PA camera in 2023, and the Samsung Galaxy S25 and Google Pixel 10 now ship with native signing capabilities, bringing credential generation to mainstream consumer hardware. LinkedIn, TikTok, and Cloudflare support or preserve these credentials.
Regulation is following suit. Enforcement of Article 50 of the EU AI Act begins in August 2026, requiring machine-readable disclosure on AI-generated content. California's SB 942 took effect in January 2026. But a critical gap exists: signing is running ahead of verification. Because most distribution intermediaries still strip embedded metadata, signed content frequently arrives at the viewer without credentials intact. C2PA is not a system that hunts for deepfake clues or fact-checks content. It is a system that marks trustworthiness. Like provenance documents that prove the authenticity of a painting, it records which hands a piece of digital content has passed through.
A World Privacy Forum report warns that C2PA is widely misunderstood. The standard is quietly laying a new media infrastructure technology layer, and this layer generates vast shareable data about creators that can be linked to commercial, governmental, and even biometric identity systems. "Who decides what counts as 'trustworthy'?" This is not a technical question. It is a question of power.
Cybersecurity has now moved beyond technical defense to become a core element of protecting digital sovereignty. Where data resides and who holds access is not a compliance issue; it is the essence of risk management. Trust becomes the competitive advantage that determines the scale of innovation. As AI accelerates complex data processing, the ability to demonstrate that this process is transparently controlled determines the stability of infrastructure.
The reality that Palantir operates systems capable of tracking and analyzing every digital trace of an individual, including emails, phone calls, financial transactions, location data, and social media activity, under contracts with over 40 governments. The reality that, as of December 2024, Palantir's market capitalization stood at $174 billion, surpassing Lockheed Martin's $121.6 billion. That software-based surveillance technology is valued higher than traditional defense contractors is an indicator of who holds the initiative in digital trust infrastructure.
Yuval Harari's concept of the "Liar's Dividend" is at work here. The more deepfake technology advances, the greater the benefit to liars. Even when genuine evidence surfaces, it can be dismissed with "that could be a deepfake too." When the foundation of trust crumbles, the whole society shakes. Yet the irony is that systems built to rebuild that trust can themselves become instruments of surveillance. This is the fundamental dilemma facing digital trust infrastructure in 2026.
4. The Weaponization of Personal Data and the Redefinition of Privacy
Every moment of your day becomes data. You wake up and check your smartphone, ride the subway to work, buy coffee at a cafe, shop online. All of it is collected and analyzed without your consent, used as the basis for decisions about you. And now that data has become a weapon aimed back at you.
Personal behavioral data is not just something collected. It serves as a tool for behavior modification, making users more dependent on platforms. This logic hides behind opaque, vague language designed to keep users from understanding how the system works. Research has shown that patterns of Facebook 'likes' alone can predict a person's political orientation with over 90% accuracy. Location data reveals daily activity patterns, the people someone meets, the places they frequent. A person moving in patterns that deviate from their norm can be classified as 'suspicious behavior' and flagged for additional surveillance.
This weaponization of data reaches an entirely different level of threat when combined with cyberattacks. According to Unit 42's 2026 Incident Response Report, identity weaknesses were implicated in nearly 90% of investigated cases. Attackers use agents to scrape LinkedIn data and generate hyper-realistic lures. When the first social engineering attempt fails, the agent tries alternative channels through self-prompting. Trend Micro's mid-2025 scan revealed more than 200 unprotected Chroma servers and over 3,000 AI components publicly exposed online, vulnerable to data theft or model poisoning.
The FBI's 2024 IC3 Report recorded 193,407 phishing complaints. Business email compromise (BEC) losses reached $2.77 billion across 21,442 cases. Total cybercrime losses hit $16.6 billion, a 33% increase year over year. Sixty-four percent of U.S. companies experienced BEC fraud in 2024, with average losses of approximately $150,000 per incident. Behind these numbers lies the sophisticated collection and analysis of personal data.
In 2013, Edward Snowden exposed the NSA's PRISM program, which accessed servers at Google, Facebook, Apple, and Microsoft directly, collecting personal emails, photos, and call records in real time. The wiretapping of mobile phones belonging to leaders of 35 nations came to light. German Chancellor Merkel had been monitored for ten years. Thirteen years have passed since Snowden's revelations, and AI technology has advanced dramatically in that time. Where the NSA's surveillance once focused on amassing vast quantities of communications data, AI now analyzes and predicts from that data in real time. AI algorithms process in minutes what would take human analysts years.
The cunning aspect of the Five Eyes intelligence alliance is how it circumvents each country's legal constraints. U.S. law restricts the government from directly surveilling its own citizens, but receiving information about Americans collected by the UK is relatively unrestricted. This effectively neutralizes each nation's privacy protections and surveillance regulations.
China's Social Credit System represents an extreme case of personal data weaponization. CCTV, facial recognition, internet usage records, shopping histories, and traffic violations are integrated to assign a behavioral score to every individual citizen. Huawei developed a surveillance camera system designed to identify Uyghurs specifically. When the system detects a Uyghur person, it automatically triggers a 'Uyghur alarm' that notifies police. Technology became an instrument of racial discrimination.
Privacy must now be redefined beyond hiding information. It is about defending against your own data being used to manipulate you. It is about preserving human agency. Palestinian poet Mosab Abu Toha was fleeing with his three-year-old child when a facial recognition AI error misclassified him as a wanted person. He was detained for two days and tortured. A machine's mistake left an indelible wound on a human life. An intelligence officer working with Israel's Lavender system testified: "As a human being, I served no role other than stamping approval."
In Cisco's 2025 Cybersecurity Readiness Index, 86% of business leaders with cyber responsibilities reported experiencing at least one AI-related incident in the previous 12 months. Seventy-eight percent of CISOs say AI-powered threats are having a "significant impact" on their organizations. Gartner projects AI spending will grow 44% in 2026 and reach $4.7 trillion by 2029. That figure dwarfs the estimated $238 billion in spending on information security and risk management solutions projected for the same year.
Just as the World Bank's Gini coefficient captures wealth inequality in a single number, the monopolization of AI technology and disparities in data access are producing a new form of power imbalance. The reality that the top 1% owns 45 to 50 percent of global wealth repeats itself in digital space as a gap between the few who hold data sovereignty and the many who do not. At the end of this arms race, where agents trade attacks and defenses without pause, we are forced to ask whether technology exists to serve humans, or whether humans have been reduced to data fuel for technology.
Kim Kyung-jin
Attorney · Former Member of the National Assembly · AI Policy Researcher
© 2026 Kim Kyung-jin. All rights reserved.