ESET has previewed a new set of AI security capabilities at GITEX AI ASIA 2026 in Singapore, targeting the growing risks enterprises face as employees adopt AI tools without IT oversight — and as agentic AI expands the attack surface beyond traditional endpoints.

The features, first demonstrated at RSAC 2026 and slated for launch later this year, will be integrated into the ESET PROTECT Platform and are built as browser security functions designed to intercept and analyse AI interactions in real time.

Scanning Prompts and Responses to Contain Shadow AI

ESET said many employees are already using open cloud chatbots without IT visibility, creating what the company describes as “shadow AI” risks — inadvertently exposing internal documents, API keys, and credentials to public AI systems. The new features address this by scanning both prompts and responses, flagging malicious URLs, detecting prompt injection attempts, and blocking uploads of sensitive data to public AI tools. Security teams will be able to surface this activity through the ESET PROTECT Platform for investigation and policy enforcement.

“As companies rely more on AI for productivity and automation, they face growing risks around sensitive data exposure, compliance violations, and misleading outputs,” said Juraj Jánošík, ESET Director of Artificial Intelligence. “Agentic AI is shifting the security battlefield back to the endpoint. ESET has spent over 30 years building leading endpoint protection powered by AI and machine learning, so we’re uniquely positioned to help organisations secure this next wave of AI right where it starts.”

AI Supply Chain Attacks on the Rise

Beyond chatbot risks, ESET warned of a rise in AI supply chain attacks — including trojanised components in widely used libraries such as LiteLLM, and autonomous agents capable of executing actions on systems with limited oversight. The company said it has been protecting customers from such attacks through compromised libraries delivered via standard repositories and is expanding its research in this area.

ESET also launched a free AI Skills Checker tool, available to non-ESET customers, which analyses AI skills for hidden instructions, malicious code, and risky behaviour using multilayered inspection and cloud-based sandboxing.

Industry Collaboration on Agentic AI Standards

ESET said it is the only dedicated cybersecurity member of the Agentic AI Foundation (AAIF), working alongside OpenAI, Amazon, Microsoft, and Anthropic to establish trusted standards and secure protocol designs for AI agent interoperability.

GITEX AI ASIA 2026 runs on 9–10 April at Singapore Expo. ESET is exhibiting at Booth HB-A106 in Hall B.