Open Source Policy Frameworks

Introducing the Model Openness Framework Abstract Generative AI (GAI) offers unprecedented possibilities but its commercialization has raised concerns about transparency, reproducibility, bias, and safety. Many "open-source" GAI models lack the necessary components for full understanding and reproduction, and some use restrictive licenses, a practice known as "openwashing." We propose the Model Openness Framework (MOF), a ranked classification system that rates machine learning models based on their completeness and openness, following principles of open science, open source, open data, and open access. The MOF requires specific components of the model development lifecycle to be included and released under appropriate open licenses. This framework aims to prevent misrepresentation of models claiming to be open, guide researchers and developers in providing all model components under permissive licenses, and help companies, academia, and hobbyists identify models that can be safely adopted without restrictions. Wide adoption of the MOF will foster a more open AI ecosystem, accelerating research, innovation, and adoption. Whitepaper (Google Doc – open for public comment): https://lnkd.in/dFkvXvHT

💡 Lately, I've been grappling with my values around #opensource and how it applies to AI. In a new Wisconsin Law Review article my colleague Chinmayi Sharma argues that framing AI as simply “open source” or “closed” misses the mark. AI systems are built from many layers — compute, models, data, APIs, monitoring — and each layer can be opened or restricted in different ways. The authors propose a “differential openness” framework: deciding which parts of AI should be open, to whom, and under what conditions. The framework "unbundles" the open-source concept for AI across key components including #hardware that powers AI, #trainingdata that shapes capabilities, #modelweights that encode knowledge, #sourcecode that defines structure, #operationalcontrols that reveals performance characteristics, and the #humans putting it all together. For policymakers, this means: 🤚 Stop asking “open or closed?” 👉 Start asking “which parts of the system should be open, and how?” Bottom line: Openness isn’t binary — it’s a policy lever. Used wisely, it can foster innovation and oversight while managing real risks. Highly recommend a read! https://lnkd.in/edEkp-rg

🤔 Weekend Reading: "Data Governance in Open Source AI: Enabling Responsible and Systematic Access" 📚 Open Source AI thrives on shared datasets. Yet, the current landscape is fraught with access challenges. Last year, I had the privilege of joining a group convened by the Open Source Initiative (OSI) and Open Future Foundation, focused on designing data governance frameworks to enable fair and responsible data access in open source AI. 🤝 💡The result of this convening, masterfully refined by Alek Tarkowski, is now available as a white paper! READ: https://lnkd.in/enxZTsTf The paper highlights two key paradigm shifts needed to better govern data for open source AI: 👉 Adopting a data commons approach: Moving beyond traditional open data frameworks toward broader data commons governance (and new types of data collaboratives). 👉 Expanding the stakeholder universe: Bringing in more voices from the AI lifecycle, including data stewards and impacted communities (to responsibly create, curate, and share new datasets) 🔑 The paper also identifies six focus areas to drive progress in Open Source AI data governance: 1️⃣ Data preparation and provenance 2️⃣ Preference signaling and social licensing 3️⃣ Data stewards 4️⃣ Environmental sustainability 5️⃣ Reciprocity and compensation 6️⃣ Policy interventions What's next? Call for collaboration among developers, policymakers, and civil society organizations to create shared standards and solutions that balance open access with responsible governance. 🌍 Colleagues who participated in the workshop that impacted the outcome include: Renata Avila, Dr. Ignatius Ezeani, Ramya Chandrasekhar, Maximilian Gahntz, Deshni Govender, Masayuki Hatta, Julie Hunter, Paul Keller, Stefano Maffulli, Ricardo Mirón Torres, Kristina Podnar, Aviya Skowron, Anna Tumadóttir, Joana Varon, Thom Vaughan, Stefano Zacchiroli, Mer Joyce #data4good #opensourceAI #datacommons #datagovernance #publicAI #opensource