OSI Establishes AI Openness Classification Framework for G7

·2026.06.02 05:44

Key point

OSI has established a four-tier model classification framework to define AI openness for the G7.

OSI (Open Source Initiative) has proposed four categories defining AI Openness to concretize the G7's vision for AI openness.

The main classification framework is as follows:

Weights Available: Only model weights are released, under a proprietary license
Open Weights: Uses an open source license, but training code and data are not released
Open Source: Training code and data are released along with an open source license (exceptions allowed when legally/technically impossible)
Open Source and Open Data: Open source models that can distribute all data

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.