ESET commits €40 million to AI security as agentic systems widen attack surface

ESET has announced a €40 million investment to accelerate its work in AI-powered cybersecurity, including security-first foundational AI models, a layered AI security stack, and a new-generation AI Security Operations Center.

The investment comes as the company warns that autonomous and agentic AI systems are creating a new attack surface for organisations. Since March 2026, ESET technologies have scanned nearly 800,000 unique AI skills, identifying about 25,000 as suspicious and blocking more than 3,000 as malicious.

AI skills become a software supply-chain concern

ESET said AI skills, modular components that instruct AI agents to perform tasks, use tools, access services, and interact with external systems, are becoming part of a fast-growing software supply-chain layer.

The number of publicly available skills has increased 13-fold from about 60,000 at the start of the year. According to ESET researchers, these skills can connect sensitive systems to external repositories, plugins, datasets, and third-party services, creating another supply-chain layer for organisations to secure.

Richard Marko, CEO of ESET, said the investment is aimed at ensuring AI strengthens cybersecurity rather than expanding risk.

“Cybersecurity is entering a completely new era,” said Marko. “Artificial intelligence is no longer only a tool for defence. It is becoming part of the attack surface itself. Our investment is focused on ensuring that AI strengthens cybersecurity rather than weakens it – and on building technologies capable of protecting organisations in a world of autonomous AI.”

The investment will also support a three-year hiring plan to grow ESET’s R&D team to 1,000 researchers and engineers.

ESET builds security-focused AI models and AI protection layers

ESET will use part of the investment to develop its own security-focused AI models, designed specifically for cybersecurity use cases. Unlike general-purpose AI systems trained on broad internet content, ESET said its models will be optimised using cybersecurity telemetry and real-world threat intelligence gathered over almost 35 years.

The company will continue expanding AI-powered technologies such as ESET LiveGrid, ESET LiveCortex, and ESET LiveGuard, while exploring emerging AI concepts such as World Models that can understand behaviour, context, and intent within digital environments.

ESET is also building a layered AI-native security architecture to protect organisations from AI-driven risks. This includes ESET Secure AI Relay, a secure intermediary layer between users, AI agents, business applications, and AI models.

The company also plans to develop network-level security protections for communication between AI agents. Its ESET AI Skills Checker, launched as a free tool at RSAC 2026 Conference, was designed for the emerging ecosystem of agentic AI systems.

AI SOC development targets alert volume and response complexity

The third part of ESET’s investment centres on a new generation of AI Security Operations Center technologies. The company said the aim is to rethink how cybersecurity telemetry is processed, correlated, and understood, rather than simply replacing analysts with AI agents.

ESET said the approach is intended to make advanced AI-powered cybersecurity more accessible to large enterprises, mid-sized companies, and smaller organisations through highly automated, supervised protection technologies.

“Cybersecurity cannot scale by adding more alerts, more dashboards, and more complexity,” Marko added. “The industry needs another major leap forward. We believe AI must help make world-class cybersecurity effortless and available to everyone.”

The company also linked the investment to technological independence and European cybersecurity sovereignty, arguing that cybersecurity-specific AI should not depend entirely on models controlled by a small number of major technology companies.