Brawli Kun (Bradley M. Kuhn), Executive Director and one of the creators of the human rights organization Software Freedom Conservancy (SFC), criticized the OSI (Open Source Initiative) Determination of the Open AI system. According to Kun, the OSI organization rushed to publish the final determination without thorough discussion, unlike the lengthy process that led to the definition of Open Source. The current definition of an open AI system should be considered a recommendation rather than a definitive term.
Kun is concerned that the current criteria for open AI could have significant repercussions, erode the credibility of Open Source, and divide the community. He argues that the OSI’s decision not to require disclosure of training data used in AI models could lead to misuse of the term “open.” OSI defended its criteria as a necessary measure to prevent misuse of the term “open” in the rapidly evolving field of AI technology.
The OSI’s determination of an open AI system requires the disclosure of training data details, but not the actual data itself. This limitation could hinder full replication of AI models, deviating from the principles of open source code. The current definition only ensures the freedom to use and distribute AI models, while neglecting the freedoms to modify and distribute modified versions.
Overall, the approved definition of an open AI system falls short of providing the comprehensive transparency and accessibility typically associated with Open Source software. The lack of full disclosure of training data poses challenges in identifying potential security vulnerabilities in AI models.