The Open Source Initiative's executive director departs - what it means for the OSAID debate (www.zdnet.com)

Stefano Maffulli, the Open Source Initiative’s first executive director, will step down in October to pursue work on open-source AI and data governance. During his 2021–2025 tenure he professionalized OSI, expanded international partnerships and shepherded the much-discussed Open Source AI Definition (OSAID) 1.0. Deborah Bryant will serve as interim director while the board searches for a successor; the OSI says it will continue emphasizing licensing, policy and building norms for open-source AI. OSAID 1.0 is a milestone because it tries to translate classical open‑source freedoms—use, study, modify, share—into the world of AI by addressing models, weights and code. Crucially, the definition compromises on training data transparency by requiring “sufficiently detailed” data disclosure rather than full open access, a concession aimed at legal and privacy constraints. That compromise has split the community: major backers (Mozilla, SUSE, Bloomberg Engineering, CNCF) see OSAID as a foundational step for licensing and transparency, while critics (including Red Hat’s Richard Fontana, Luca Antiga, and others) argue it leaves loopholes around weights and datasets that vendors could exploit. The debate matters because OSAID will shape what “open source AI” legally and practically means—impacting adoption, interoperability, and how companies label models like Llama, Grok or Phi-2. The next OSI leader will need to navigate these technical, legal and ethical trade-offs as the definition evolves.