I keynote and do talks on open source AI and open models, the agent stack, AI and agentic security, and AI in finance — pitched anywhere along the range, from deeply technical to the policy, economic and societal consequences of building AI in the open. Below: the topics I cover, bios in three lengths, press photos and how to book.
What “open” has to mean for a model release to be auditable and reproducible — the Model Openness Framework, the OpenMDW license, and why so much of what is called open source AI is not.
Why the security surface of an agent system is categorically harder than model safety — runtime tracing, evaluation, and the standards that have to exist before agents can be trusted in production.
Open financial language models, benchmarking and FinAI governance — from the Open FinLLM Leaderboard to privacy-preserving training and agents that act in markets.
Matt White is Founding Executive Director of the Open Intelligence Foundation and a Visiting Scholar at Columbia University. He is the former Global CTO of AI at the Linux Foundation, former Executive Director and CTO of the PyTorch Foundation and the founding CTO of the Agentic AI Foundation.
Matt White is Founding Executive Director of the Open Intelligence Foundation and a Visiting Scholar at Columbia University. He is the former Global CTO of AI at the Linux Foundation, former Executive Director and CTO of the PyTorch Foundation and the founding CTO of the Agentic AI Foundation. He founded the Generative AI Commons, and was lead author of the Model Openness Framework and the OpenMDW license.
Matt White is Founding Executive Director of the Open Intelligence Foundation, which was founded to protect and advance open source AI. It runs several programs and hosts Open Source AI Week. He is also a Visiting Scholar at Columbia University, working on agentic AI research for financial intelligence.
Over his career he has built the platforms, licenses and governance that make open artificial intelligence practical — as former Global CTO of AI at the Linux Foundation, former Executive Director and CTO of the PyTorch Foundation, founding CTO of the Agentic AI Foundation and Director of the Generative AI Commons. He was first author of the Model Openness Framework and the architect of the OpenMDW license, the first model-specific permissive license, used by NVIDIA and others.
Previously Matt was Head of AI and Data at Amdocs and National Director of Data Science at IBM. He holds an M.Sc. in Data Science from UC Berkeley, an MBA from the University of Denver and a B.Sc. in Information Technology from York University.
High-resolution, free to use with credit.
