Binary Classification of AI Models as Open or Closed-Source

Determine whether AI models can or should be dichotomously classified as open-source or closed-source, and establish precise criteria that justify such a binary classification for AI systems.

Background

The paper reviews the rapid rise of “open models” and highlights ongoing debates about what constitutes openness in AI. It notes that many projects billed as open source do not meet OSI standards and that companies have been criticized for “open-washing.”

The Open Source Initiative (OSI) is actively developing a definition of open source AI, while other scholars propose a gradient of openness rather than a binary distinction. Against this backdrop, the authors explicitly state that whether AI models should be classified as either open or closed-source remains unresolved.