At RSAC 2026, ESET unveiled an innovative AI protection feature designed to identify malicious URLs through chatbot interactions. This system logs activity at the endpoint and allows for thorough investigation using the ESET PROTECT Platform.
ESET, a leading name in cybersecurity, has announced the introduction of advanced AI protection capabilities aimed at enhancing the security of employee interactions with AI tools. Showcased at RSAC 2026 and expected to launch later this year, these features will increase visibility within the ESET PROTECT Platform, allowing for a thorough examination of emerging threats related to everyday AI usage and the adoption of agentic AI across enterprises.
“As companies increasingly depend on AI for productivity and automation, they are confronted with heightened risks, such as sensitive data exposure, compliance breaches, and erroneous outputs. The rise of agentic AI is transforming the security landscape, demanding attention on the endpoint. With over 30 years of expertise in AI and machine learning, ESET is uniquely positioned to aid organizations in safeguarding this new wave of AI from its origin.”
– Juraj Jánošík, ESET Director, Artificial Intelligence
With the integration of AI tools into daily workflows, many employees are turning to cloud-based chatbots without adequate IT oversight, which introduces “shadow AI” risks that could expose sensitive information like internal documents, API keys, and other credentials. ESET combats these risks with a suite of technologies designed to monitor interactions closely, including secure browser technology that analyzes AI prompts and responses in real-time. This proactive approach helps prevent data leakage and identifies harmful or misleading content before it affects users.
During its demonstration at RSAC 2026, ESET’s new AI protection feature successfully flagged malicious URLs from chatbot submissions, capturing endpoint activity and making it accessible in the ESET PROTECT Platform for investigation. This same technique is applied to detect prompt injection attempts, scripts, and sensitive data entries, enabling organizations to either block or monitor activities compliant with their policies. Security teams will benefit from enhanced visibility into AI tool usage through ESET PROTECT Platform logging, which aids in risk assessment and policy enforcement.
As organizations enhance their implementation of agentic AI tools, potential attack vectors expand beyond chatbot interactions, introducing risks associated with the AI supply chain. This includes compromised AI frameworks and tools, such as trojanized components in commonly used libraries like LiteLLM, along with autonomous agents like OpenClaw that perform actions within systems with minimal oversight. ESET has been defending clients against supply chain attacks linked to compromised libraries from standard repositories and is noting a concerning trend in these attacks, reinforcing its commitment to ongoing research and development in relation to AI tools.
In line with its broader innovation strategy in AI security, ESET launched a complimentary ESET AI Skills Checker at RSAC 2026. This tool, which is accessible to non-ESET customers and built using the same technology as ESET’s endpoint security products and ESET LiveGuard, evaluates AI skills for hidden instructions, malicious code, and risky behaviors through multi-layered inspection and cloud-based sandboxing. It is currently a built-in feature for existing ESET Endpoint users.
For over 30 years, ESET has been at the forefront of lightweight, high-performance endpoint security powered by machine learning and artificial intelligence. These new capabilities enhance that foundation, equipping organizations to face today’s rapidly evolving threat landscape, where cybercriminals increasingly leverage AI to scale their attacks, target employees, and orchestrate sophisticated social engineering schemes.
As a dedicated cybersecurity member of the Agentic AI Foundation (AAIF), ESET is committed to securing emerging AI agent communication protocols by collaborating with industry leaders such as OpenAI, Amazon, Microsoft, and Anthropic. Together, they aim to establish trusted standards, create secure protocol designs, and define best practices for interoperability among AI agents.
