ESET unveils AI security tools for business agents

ESET has unveiled three AI security products for organisations, extending its security portfolio to cover AI agents and generative AI use.

The new products - ESET AI Agent Security, ESET AI Behavioural Monitoring and ESET AI Conversation Security - were introduced as part of a broader push into AI-related threat detection, investigations, threat protection and security operations.

ESET AI Agent Security focuses on the software components that AI systems rely on. It is designed to inspect AI-related components across what ESET describes as the AI supply chain, including files, downloaded elements, external services, repositories, code interactions, skills and plugins used by autonomous agents.

ESET AI Behavioural Monitoring is aimed at the actions of AI agents once they are operating inside an environment. It is intended to detect and block suspicious or malicious behaviour, including attempts to access inappropriate resources, run unsafe actions or move beyond the expected scope of a task.

ESET AI Conversation Security targets the use of generative AI tools by employees and other users. It inspects uploaded files and metadata for sensitive content to reduce accidental data leakage and unsafe prompting, and flags responses linked to phishing or malicious and unwanted websites.

The launch reflects a shift in security concerns as businesses adopt AI tools and autonomous agents, sometimes without direct oversight from IT teams. That trend has contributed to what security providers describe as shadow AI, where systems are used inside organisations without formal governance or visibility.

ESET said its research had identified thousands of malicious AI skills running on tools such as OpenClaw. According to the company, those skills could allow attackers to gain access to systems, remove data, download and execute malware, and override user instructions.

AI oversight

Kamil Pšenák, Senior AI Product Manager at ESET, framed the issue as treating AI systems as active participants inside a company's estate rather than passive tools.

"AI is a new class of actor inside the company - reading, writing, making decisions and executing. As such, it deserves the same security attention as users and endpoints, because it behaves like both at once," said Kamil Pšenák, Senior AI Product Manager at ESET.

He also linked the rise of AI use to a renewed focus on endpoint security.

"It is all converging around the endpoint. What used to be a neglected part of the cybersecurity stack is now primed to shine anew thanks to the proliferation of AI tools at a business level," Pšenák said.

The three features will be available across the ESET PROTECT portfolio. Rather than being sold as separate add-ons, they are being integrated into the existing platform alongside the company's AI technologies and detection systems.

That approach matters for customers and resellers because many organisations are already managing multiple security tools across endpoints, cloud systems, identity controls and email. Folding AI-related controls into an established management layer could make it easier for IT teams to apply policy and monitor use without adding another standalone console.

It also points to a commercial opening for channel partners, which are increasingly being asked by customers how to manage the risks tied to large language models, AI assistants and autonomous agents in day-to-day business processes.

Ryan Grant, Country Manager, U.S. and Canada at ESET, said discussions with partners were changing as AI use spread through customer environments.

"AI is changing the conversations our partners are having with customers," said Ryan Grant, Country Manager, U.S. and Canada at ESET.

"Instead of simply reacting to new AI risks, partners can lead with a prevention-first strategy that helps businesses adopt AI securely while reducing operational complexity. That's a meaningful opportunity to deepen customer relationships and grow recurring security services," Grant said.

ESET has used artificial intelligence and machine learning in its products for more than two decades, according to the company. The latest launch shows how cybersecurity suppliers are now applying those same techniques to securing AI systems themselves, as organisations move from experimental use of chatbots and assistants to broader adoption of autonomous software agents.