AI-Driven Cyberattacks Accelerate as Breakout Times Shrink

Recent findings from CrowdStrike’s 2026 Global Threat Report highlight a marked increase in the use of artificial intelligence within cyberattacks, with organizations facing faster and more complex threats.

The report, released on February 24, 2026, details how threat actors are leveraging AI to enhance their capabilities in areas such as credential theft, data exfiltration, and evasion techniques. According to CrowdStrike, cybercriminals have adopted new strategies that involve injecting harmful prompts into generative AI tools across dozens of organizations.

Malicious actors are not only using AI to automate attacks but are also exploiting vulnerabilities within AI development platforms. These tactics include establishing persistent access, deploying ransomware, and setting up deceptive AI servers that mimic legitimate services to intercept confidential information.

CrowdStrike’s data shows that the speed of cyber intrusions has increased, with the average time for eCrime actors to move from initial access to broader network compromise now at 29 minutes. In some cases, data exfiltration has started within just four minutes of gaining entry.

What the numbers show

• AI-enabled cyberattacks increased by 89% year-over-year
• The fastest observed eCrime breakout occurred in 27 seconds
• In one case, data exfiltration began within four minutes of access

More than 90 organizations have experienced cyber incidents involving the misuse of generative AI tools. Attackers have used these tools to generate commands aimed at stealing credentials and cryptocurrency, demonstrating a broadening threat landscape.

Threat actors have also published malicious AI servers that impersonate trusted platforms, allowing them to intercept sensitive data from unsuspecting users. These methods indicate an evolving approach to cybercrime, with AI systems themselves now becoming part of the attack surface.

CrowdStrike CEO George Kurtz stated that as AI is integrated into development pipelines, SaaS platforms, and operational workflows, the systems themselves present new security challenges. Kurtz also said that this trend increases the necessity for robust cybersecurity measures to address the expanding attack surface created by AI adoption.

* This article is based on publicly available information at the time of writing.