The 2026 Global Threat Report released by CrowdStrike outlines how AI is accelerating adversary capabilities and expanding the enterprise attack surface. In 2025, the average eCrime breakout time fell to 29 minutes, with the fastest observed breakout occurring in 27 seconds.
Adversaries are using AI to target organisations, including injecting malicious prompts into generative AI tools at more than 90 organisations and exploiting vulnerabilities in AI development platforms. The report notes that as technological innovation advances, adversaries continue to adapt their methods to take advantage of new systems.
According to the analysis, AI-enabled adversaries increased their activity by 89% year over year, using AI for reconnaissance, credential theft and evasion. Intrusions are occurring more quickly and are increasingly difficult to detect, moving through trusted identities, SaaS applications and cloud infrastructure while reducing the time available for defenders to respond. In this environment, AI is both a tool used by attackers and a target in its own right.
Key findings from the report include:
- AI as an Attack Surface: Adversaries injected malicious prompts into legitimate generative AI tools at more than 90 organisations to generate commands for credential theft and cryptocurrency theft. They also exploited vulnerabilities in AI development platforms to establish persistence and deploy ransomware, and created malicious AI servers impersonating trusted services to intercept sensitive data.
- Record Breakout Times: The average eCrime breakout time fell to 29 minutes, representing a 65% increase in speed compared with 2024. The fastest observed breakout occurred in 27 seconds. In one case, data exfiltration began within four minutes of initial access.
- Nation-State and eCrime Activity: AI-enabled adversaries increased operations by 89%. Russia-nexus actor FANCY BEAR deployed LLM-enabled malware known as LAMEHUG to automate reconnaissance and document collection. eCrime actor PUNK SPIDER used AI-generated scripts to accelerate credential dumping and remove forensic evidence. DPRK-nexus actor FAMOUS CHOLLIMA used AI-generated personas to scale insider operations. China-nexus activity increased by 38% in 2025, with 67% of exploited vulnerabilities delivering immediate system access and 40% targeting internet-facing edge devices. DPRK-linked incidents rose by more than 130%, and PRESSURE CHOLLIMA’s $1.46 billion cryptocurrency theft was the largest single financial theft reported.
- Zero-Day and Cloud Exploitation: Forty-two per cent of vulnerabilities were exploited before public disclosure. Cloud-focused intrusions increased by 37% overall, including a 266% rise in activity from state-linked threat actors targeting cloud environments for intelligence collection.
Adam Meyers, head of counter adversary operations at CrowdStrike, said that breakout time reflects how quickly intrusion methods are evolving, and that security teams must operate at greater speed to respond effectively to modern threats.
