🎉Our Lab Position Paper on Open-Weight Foundation Model Governance Accepted to ICML 2026

Paper: Position: Current Model Cards Are Insufficient for Downstream Governance of Open-Weight Foundation Models

Authors: Sungwon Chae, Keonwoo Kim, Hoki Kim, Jaeyeon Ju, Sangchul Park

Venue: ICML 2026

Our lab position paper “Current Model Cards Are Insufficient for Downstream Governance of Open-Weight Foundation Models” has been accepted to ICML 2026, one of the most prestigious venues in machine learning.

As open-weight foundation models proliferate rapidly, the limits of existing governance mechanisms for controlling downstream risks are becoming increasingly clear. Model cards, currently treated as the de facto standard, document a model’s training data, performance, and limitations — but they fall short as a tool for enforcing responsibility and authority at the point of use.

This paper argues directly that model cards alone are insufficient for the downstream governance of open-weight models. As an alternative, the authors propose an integrated governance framework that combines model cards (transparency), use policies (behavioral norms), and licenses (legal enforcement), demonstrating that responsible AI deployment requires policy, law, and technology working in concert.

The paper will be presented at ICML 2026.