With the exponential advancement of artificial intelligence (AI) technology, highly sophisticated deepfakes and fabricated information have emerged as humanity's top new threat. Experts warn that the convergence of generative AI and autonomous agent technologies has evolved far beyond simple internet scams or cyber hacking, shaking the foundational trust of society and disrupting the global economic system entirely.
According to the World Economic Forum's (WEF) 'Global Risks Report 2025', nearly 1,500 experts from business, government, and academia identified 'misinformation and disinformation' as the most severe threat facing the world over the next two years. This outranked extreme weather events, societal polarization, conventional cyberattacks, and interstate armed conflicts. Dave Aron, Distinguished VP Analyst at Gartner, explained that "disinformation is a 'meta-issue' because it systematically erodes our collective capacity to address all other critical global problems."
Evolving Deepfakes and Surging Corporate Financial Losses
Generative AI-driven deepfakes are already inflicting catastrophic financial damage on the business sector. In a prominent case, a branch employee at the British engineering firm Arup in Hong Kong was deceived by a deepfake video conference where fraudsters perfectly mimicked the company's CFO and other high-ranking executives. Convinced the meeting was real, the employee transferred $25 million to fraudulent accounts. Global economic losses stemming from digital disinformation reached an estimated $78 billion as early as 2019, and by 2028, corporations are projected to spend over $30 billion annually to combat disinformation—amounting to roughly 10% of their total marketing and cybersecurity budgets.
Amplified Threats via 'Agentic AI' and Automated Mass Distribution
What alarms experts the most is the integration of disinformation with 'Agentic AI'—systems that autonomously execute complex goals. If a malicious actor commissions a disinformation campaign, Agentic AI can independently generate deceptive videos, forged audio evidence, and personalized narratives, automatically distributing them to over a million targeted individuals simultaneously. The speed of such AI-driven manipulation completely outpaces human detection and traditional incident response. Furthermore, new vulnerabilities like 'Model Poisoning'—the intentional injection of corrupted or biased data into AI training sets to distort the model's core logic—are expanding the cybersecurity battleground.
Prebunking as the Strongest Vaccine Amid Election Risks
A recent survey revealed that while 79% of corporations experienced disinformation incidents over the past three years, only 38% have established a formal response framework. To address this, Gartner proposes the 'TrustOps' framework, which consists of Debunking (tracking and refuting false claims), Prebunking (proactively educating stakeholders), and Grounding (verifying data authenticity).
Aron emphasized that 'prebunking'—educating individuals on manipulation techniques before they encounter fake content—serves as the most effective preemptive vaccine. With the upcoming June 3 local elections in South Korea stoking intense anxieties over deepfake-driven election interference, public and regulatory vigilance must be elevated to its highest level during seasons of critical democratic decision-making.
[Copyright (c) Global Economic Times. All Rights Reserved.]
