Introduction
The Responsible AI Workstream (part of the Generative AI Commons) has completed the Responsible Generative AI Framework (RGAF) document (DOWNLOAD IT HERE). The RGAF is intended to help AI systems implementers and consumers to navigate the complex responsible AI challenges of open source generative AI projects. The RGAF identifies nine dimensions relevant to responsible AI together with challenges and remediations. The nine dimensions of RGAF align with the elements from major global AI frameworks such as the EU AI Act, NIST AI Framework, Singapore AI Strategy, and China’s AI Development. The RGAF describes how the Model Openness Framework (MOF) and the Decoding Trust initiatives relate to the nine dimensions, and includes an outline of how RGAF relates to AI systems as a whole.
What’s Happened so Far
The Responsible AI Workstream started out as the Trusted AI Committee in the LF-AI and Data. The following is an outline timeline for the development of the RGAF:
- January–March 2024 : Reformulated the Trusted AI Committee as the Responsible AI Workstream at the request of the Generative AI Commons
- February–November 2024 : Produced RGAF
- December 11 2024-January 5 2025: Reviewed by RGAF community – in parallel companion slides were created
- January 5-January 22 2025 : Incorporated RGAF community comments
- January 22 2025 : Reviewed by Generative AI Commons
- January 30 2025 : Reviewed by LF-AI and Data TAC (Technical Advisory Council) – agreement to publish received
- February-March 2025: Reviewed by LF-AI and Data leadership, finalized, and published
The RGAF was presented at a meetup in NYC on March 5, 2025 to an animated audience of nearly 200 people, and more outreach sessions are planned. The FAQ at the end of this blog is a synthesis of some of the many questions received at the meetup.
What’s Next?
We are inviting interested parties to join us in developing the Responsible AI workstream on Pathways
Please join the Responsible AI workstream here.
Some FAQ
Would it be possible to have a shorter version of RGAF for startups?
Join the Responsible AI Workstream to work on a shorter version.
Who is the main target of the RGAF?
Generative AI Systems implementers and consumers, as well people with a casual interest in responsible AI.
Do we want the RGAF to become regulations?
The RGAF is intended to provide a set of helpful guidelines and there is no plan to aim for regulations.
How does the political climate impact the work?
The RGAF is focused on providing helpful guidelines and is a global community driven effort.
Can we work with you on RGAF?
Yes, please join the Responsible AI workstream here
