Beyond Open Weights: How Open Source and Open Standards Are Shaping the Future of AI

As AI becomes increasingly foundational to software, science, and society, the open source community faces new challenges—and unprecedented opportunities. In this talk, Matt White, Executive Director of the PyTorch Foundation and GM of AI at the Linux Foundation, will share how open source principles are evolving to meet the complexities of modern AI.

The session will trace the trajectory from "open weights" models to truly permissive open models, and examine how frameworks like the Model Openness Framework and model-specific licenses like OpenMDW are helping developers and organizations evaluate the completeness and usability of AI models.

Matt will also address the growing importance of open standards in supporting the next wave of AI systems—particularly agentic, embodied, and autonomous AI. He will unpack the practical implications of legislative trends, licensing structures, and model governance, and share actionable strategies for developers, researchers, and enterprises seeking to build responsibly in this fast-changing environment.

Attendees will walk away with a clear understanding of:

The difference between open weights and permissively licensed open models

How to assess model completeness and openness and what they mean for usability

Why open standards are essential for interoperability and innovation

How to develop open source projects that will stand the test of time

As the open source ecosystem moves beyond releasing code to shaping foundational infrastructure, this talk offers a roadmap for where open AI is headed—and how to help shape it.