Cybersecurity researchers have confirmed what many in the industry feared: artificial intelligence is now playing a substantial role in creating the malware that threatens businesses and individuals worldwide.

According to experts at Kaspersky, one of the world's leading cybersecurity firms, large language models — the same technology that powers tools like ChatGPT — are being used to generate significant portions of malicious software. The finding, reported by Gadget, marks a troubling milestone in the ongoing arms race between cybercriminals and security professionals.

From Theory to Practice

The use of AI in cyberattacks has long been discussed in security circles as a theoretical concern. That concern is now reality. While major AI companies have implemented safeguards to prevent their models from assisting with malicious activities, determined actors have found ways around these restrictions or have developed their own uncensored models.

The implications are significant. Malware development has traditionally required specialized programming knowledge and time-intensive coding work. Large language models can dramatically accelerate this process, allowing even less technically sophisticated criminals to generate functional malicious code. They can also help experienced developers work faster, test variations more quickly, and adapt their tools to evade detection.

Kaspersky's researchers did not specify what percentage of malware now contains AI-generated code, nor did they detail which types of threats are most commonly being developed this way. However, the fact that the phenomenon is now substantial enough to warrant public reporting from a major security firm suggests the trend has moved beyond isolated experiments.

A Shifting Threat Landscape

The integration of AI into malware development changes several fundamental aspects of the cybersecurity landscape. Traditional signature-based detection systems rely on identifying known patterns in malicious code. When AI can rapidly generate variations of malware with different code structures but identical functions, these detection methods become less effective.

Security teams have long relied on the fact that developing sophisticated malware requires time and expertise — factors that naturally limited the volume and variety of threats. AI removes many of these constraints. A single malicious actor could potentially generate dozens of malware variants in the time it would have previously taken to hand-code one.

The technology also lowers barriers to entry. Cybercrime has already become increasingly accessible through "malware-as-a-service" platforms, where criminals rent access to tools and infrastructure. AI-assisted development accelerates this democratization, potentially expanding the pool of active threat actors.

The Defense Response

Cybersecurity firms are not standing still. Many are deploying their own AI systems to detect and respond to threats more quickly. Machine learning models can analyze patterns across millions of files and network events, identifying suspicious behavior that might escape human analysts or traditional rule-based systems.

This creates a new dynamic: AI-powered attacks versus AI-powered defenses. The outcome of this technological arms race will depend not just on the sophistication of the models involved, but on the resources, data, and expertise available to each side.

Kaspersky and other security firms have access to vast datasets of malware samples and attack patterns, which they can use to train defensive AI systems. However, attackers have the advantage of initiative — they can test their AI-generated malware in private until it successfully evades detection, then deploy it in the wild.

Broader Implications

The use of AI in malware development is part of a larger pattern in which powerful technologies become tools in both legitimate and criminal hands. The same large language models that help programmers write code faster, assist students with homework, and enable new creative applications are now being weaponized.

This dual-use nature of AI technology presents ongoing challenges for developers and policymakers. Overly restrictive controls might hamper beneficial uses, while too little oversight could accelerate harmful applications. The technology itself is neutral; its impact depends entirely on how it is deployed.

For businesses and individuals, the rise of AI-generated malware reinforces the importance of layered security approaches. No single defense is sufficient when threats are evolving this rapidly. Regular software updates, employee training, network monitoring, and incident response planning all remain essential components of effective cybersecurity.

The Kaspersky findings serve as a reminder that the cybersecurity field is entering a new era. The tools available to both attackers and defenders are becoming more sophisticated, and the pace of change is accelerating. Organizations that fail to adapt their security strategies to this new reality do so at considerable risk.