Attack  ·  Glossary

Dependency confusion attack

An attacker registers a package on a public software library repository (like PyPI or npm) using the same name as a private, internal package used by a real project. When the build system automatically downloads dependencies, it picks up the attacker's malicious version instead of the legitimate internal one — silently injecting malicious code into the software during the build process.
AI model serving infrastructure — including the official Docker images for widely deployed LLM engines — can be quietly backdoored at build time through this technique, with no obvious sign of compromise. Any organisation building or deploying AI infrastructure from public package repositories is potentially exposed.
References
MITRE ATLAS — AML.T0010 ML Supply Chain Compromise
Track this in the live feed See how this plays out in real AI security and governance developments.
Open the feed →