ESET Unveils AI Security Features to Protect Chatbots, Workflows, and Agentic Systems

ESET has previewed a new suite of AI security capabilities designed to protect enterprise use of chatbots, AI workflows, and agentic systems, showcased at the RSA Conference 2026.

The upcoming features focus on securing the full lifecycle of AI interactions by scanning both user prompts and AI-generated responses in real time. Delivered as a browser-based security layer, the solution is designed to block malicious links, scripts, and harmful content generated by large language models, while also preventing the upload of sensitive or confidential data into public AI tools.

As AI adoption grows across workplaces, organizations are increasingly facing risks related to “shadow AI,” where employees use external AI tools without oversight. According to Juraj Jánošík, this creates exposure to data leaks, compliance violations, and misleading outputs. ESET’s approach addresses these risks at the endpoint level, intercepting AI interactions and analyzing them before they can cause harm.

The new capabilities also extend to securing personal AI assistants and defending against emerging AI supply chain threats, including compromised frameworks and malicious components embedded in widely used AI libraries. ESET demonstrated how its system can detect prompt injection attacks, flag suspicious URLs, and provide visibility into AI usage through its PROTECT Platform.

In addition, ESET introduced a free AI Skills Checker tool, which scans AI “skills” or integrations for hidden instructions, malicious code, and risky behavior. Built on the same technology as ESET’s endpoint security solutions, the tool uses multi-layered analysis and cloud-based sandboxing to identify potential threats before deployment.

The company is also collaborating with industry leaders such as OpenAI, Amazon, Microsoft, and Anthropic through the Agentic AI Foundation to establish secure standards and protocols for AI agent communication.

With these developments, ESET aims to extend its long-standing expertise in endpoint security into the rapidly evolving AI landscape, helping organizations manage new risks as AI becomes deeply embedded in everyday business operations.