Professor warns AI agents widen software supply chain risk
At the KIISC 2026 AI x RMF Workshop, a Kyonggi University professor said AI-BOM should anchor automated supply chain checks in CI/CD.
A Kyonggi University professor warned on October 22, 2026 that AI agents are widening the software supply chain's attack surface, as they now choose open-source packages themselves. The professor spoke at the KIISC 2026 AI x RMF Workshop, held at Samsung SDS West Campus in Seoul, according to dailysecu.com.
In a talk titled "Attack Surface of the Agentic AI Supply Chain and the AI-BOM Automated Verification System," the professor said that risk can spread across dependencies, training data, models and cloud access once agents select components on their own.
He cited LiteLLM as an example, noting it records about 97 million monthly downloads and 1,705 dependency packages. He also named Trivy and LiteLLM among open-source components involved in supply chain incidents, and described a "blast radius" in which compromising a single package or model can reach development tools, deployment systems and cloud infrastructure.
His proposed starting point is AI-BOM (AI Bill of Materials), an extension of the software bill of materials that documents AI supply chain components. He stressed that AI-BOM is not itself a security solution but a basis for transparency and risk assessment, and called for automated verification throughout the CI/CD pipeline.
Regulatory pressure for machine-readable evidence
The professor pointed to the EU AI Act, the Cyber Resilience Act, US federal IT procurement rules and FedRAMP as moving toward machine-readable security evidence. In background remarks, he cited LAION-5B, a dataset used to train AI models, as containing inappropriate material, an example he used to argue for tracking data provenance.
Quick answers
What is AI-BOM?
AI-BOM (AI Bill of Materials) is an extension of the software bill of materials that documents AI supply chain components such as dependencies, training data, models and cloud access. The speaker said it is a basis for transparency and risk assessment, not a security solution in itself.
Where was the 2026 AI x RMF Workshop held?
At Samsung SDS West Campus in Seoul, on October 22, 2026, hosted by the KIISC.