China-linked hackers were responsible for more than 58% of all state-sponsored targeted intrusions against technology companies in the 12 months ending March 31, 2026, with artificial intelligence assets as the primary objective. The finding comes from CrowdStrike's 2026 Technology Threat Landscape Report, released June 9, and underscores a critical shift in how nation-states are competing for technological dominance.

Why Are AI Systems Becoming the Primary Target for State-Sponsored Hackers?

The scope of China's targeting spans frontier AI laboratories and smaller domain-specific model developers, according to Adam Meyers, CrowdStrike's senior vice president for counter adversary operations. "There is an AI arms race occurring between the U.S. and China, and China intends to achieve global dominance by 2030," Meyers stated. This finding aligns with a White House Office of Science and Technology Policy memo from April 23, which accused China-based entities of conducting "deliberate, industrial-scale campaigns" to distill U.S.-developed AI models by repeatedly querying them to extract capabilities and train cheaper domestic replicas.

The targeting of AI systems represents a fundamental shift in espionage tactics. Rather than stealing traditional military or financial secrets, nation-states now recognize that controlling advanced AI technology could determine geopolitical influence for decades. This reflects a broader understanding that AI capabilities, not conventional weapons, may define the next era of global competition.

How Are North Korean Actors Complementing China's AI Espionage Strategy?

While China focuses on AI model theft, North Korean cyber actors pose a parallel threat through sheer volume and innovation in social engineering. The Famous Chollima hacking group accounted for 47% of all government-linked hands-on-keyboard intrusions targeting technology firms, embedding operatives through AI-generated deepfake identities and fraudulent credentials to secure remote IT employment.

This tactic reveals how adversaries are weaponizing AI-generated content to bypass traditional identity verification. By creating synthetic identities backed by deepfake videos or cloned voices, attackers can infiltrate organizations at the human level, gaining trusted access that no firewall can prevent. North Korean cyber actors stole $2 billion in digital assets during 2025, including $1.46 billion taken from cryptocurrency exchange Bybit in the largest crypto theft on record, demonstrating the financial scale of these operations.

What Are Organizations Missing in Their Defense Against AI-Enabled Threats?

A separate 2026 Digital Risk Report from Cybersecurity Insiders, based on responses from more than 1,100 cybersecurity, fraud, risk, and trust leaders, reveals a critical vulnerability: organizations are struggling to keep pace with rapidly evolving digital threats where attackers increasingly target trust, identity, and online reputation. The research found that 84% of organizations experienced a material digital risk incident during the past year, yet only 7% of respondents consider their digital risk programs to be mature or leading.

The gap between threat exposure and defensive maturity suggests that traditional security investments in endpoints, identities, cloud environments, networks, and email systems are insufficient. Adversaries have shifted focus to the public internet, where brands, executives, employees, customers, and business workflows remain vulnerable to coordinated, multifaceted campaigns.

Steps to Strengthen Defenses Against AI-Enabled Cyber Threats

• Enhance Patching and Vulnerability Management: Organizations should take prompt action to address known vulnerabilities and implement adequate policies for urgent and critical fixes outside routine patching cycles, especially for business-critical components. Allocate sufficient resources to handle potential surges in patching demands, as frontier AI models can identify and exploit zero-day vulnerabilities at unprecedented speed.
• Implement Zero-Trust Architecture: Design system controls on the assumption that any user, device, privileged account, or network component may be compromised. Enforce least-privilege access to all business-critical components, enhance firewalls and network segmentation, and treat external and untrusted inputs as potentially adversarial.
• Establish Centralized Digital Risk Ownership: Many organizations distribute digital risk responsibility across multiple departments, creating accountability gaps. Establish a single team with end-to-end responsibility for digital risk management spanning channels, artifacts, and organizational functions to enable faster response and better coordination.
• Monitor and Protect Human Attack Surfaces: More than half of surveyed organizations reported incidents involving executive or employee impersonation during the past year. Implement comprehensive person-of-interest monitoring, threat profiling, and personal information removal programs, extending protection beyond executives to employees with privileged access to financial systems, administrative controls, or critical business processes.
• Develop AI-Generated Content Detection Capabilities: Nearly half of organizations reported confirmed or suspected synthetic-media impersonation incidents involving deepfake videos, cloned voices, or other AI-generated content. Shift detection efforts earlier in the attack lifecycle by identifying campaign infrastructure before fraudulent content reaches intended targets.

Hong Kong regulators have already begun mandating these defenses. On June 2, 2026, the Securities and Futures Commission (SFC) and Hong Kong Monetary Authority (HKMA) issued circulars to licensed firms and authorized institutions, reminding them to review and enhance cybersecurity measures in light of heightened risks posed by AI-enabled cyberattacks. This regulatory pressure reflects a recognition that frontier AI models mark a qualitative shift in the cyber threat landscape, enabling threat actors to plan and execute complex, multi-step actions autonomously and identify security flaws that have remained undetected by software developers.

The research also highlights an emerging "AI Trust Gap." Many organizations are deploying AI-powered systems to support communications, research, transactions, and decision-making processes, but visibility and control over these systems remain limited. Indirect prompt injection attacks, where adversaries embed malicious instructions within external content that AI agents consume, represent a growing concern. Only a small percentage of organizations reported having comprehensive visibility and active controls governing AI agents' external interactions, and very few have established automated containment mechanisms capable of stopping compromised agents in real time.

The convergence of state-sponsored AI espionage, criminal deepfake campaigns, and organizational blind spots creates a perfect storm for cyber risk. As frontier AI models become more capable and accessible, the technical barrier to executing sophisticated attacks continues to lower, enabling a wider range of threat actors to cause significant damage. Organizations that fail to adopt zero-trust architecture, centralize digital risk governance, and invest in AI-generated content detection will find themselves increasingly exposed to both nation-state and criminal threats in the coming years.