SOURCE-LINKED INTELLIGENCE

Why Ghost Outputs Teach: A Kernel-Based Understanding of Subliminal Learning

arXiv · AI, language, vision and robotics · article · Sep 20, 2026 · UTC

Subliminal Learning (SL) is a recently identified phenomenon in which a student model acquires downstream task capabilities by matching seemingly unrelated auxiliary outputs from a teacher, despite never observing task labels, task-specific outputs, or the original training data. While recent studies have identified where subliminal signals may reside, the optimization mechanism underlying this phenomenon remains poorly understood. In this work, we provide a mechanistic understanding of SL through the lens of learning dynamics. Specifically, we derive a chained cross-task kernel that explicitl

Read original source ↗ Open in workspace

recordType

paper

region

Global