[License-discuss] Pre-Advice for ModelGo Licenses Review

["D 莫名 (dewitt)"]

Hi McCoy,


Thanks for your reply. The MG-*-OS License requires providing a copy of the code but does not enforce a specific license for it. My lawyer opposed adding such enforcement due to potential incompatibility between the code license and our MG license.

BTW, from my understanding, the current OSAID applies to AI systems, correct? Is it a requirement for open-source model licenses to comply with it? For example, should an open-source model license include terms ensuring that its code and data are also governed by an OSI-approved license?


Best,
Moming


> On 11 Feb 2025, at 1:23 AM, McCoy Smith <mccoy at lexpan.law> wrote:
> 
> you definitely should not submit any license that doesn't comply with the OSD, like anything that is -NC.
> 
> wrt the -OS license, if it allows for only open weights but not open code, that probably violates OSD too.
> 
> On 2/9/2025 11:15 PM, "D 莫名 (dewitt)" wrote:
>> Dear All,
>> 
>> 
>> I am a researcher at National University of Singapore (NUS) working on AI licensing. Recently, I collaborated with legal experts to draft a public-use, model-specific license aimed at promoting more standardized model publishing. As you may know, there are far fewer model licenses than software licenses for general model publishing, leading many to use OSS and Creative Commons (CC) licenses instead. However, these licenses do not fully address ML-specific behaviors, potentially causing legal uncertainties (we have conducted research on this issue). 
>> 
>> To address this gap, we drafted ModelGo Licenses, a CC-style licensing framework designed specifically for models. There are nine variations of ModelGo (MG) Licenses, such as MG-BY, MG-BY-RAI, and MG-BY-OS, similar to the structure of CC licenses. Currently, we are preparing to enter the OSI license review process and start a discussion, as suggested.
>> 
>> 1) My first question is: Should I submit all variations? Since we have nine licenses that share a large portion of their terms and conditions, I wonder if submitting them all is necessary.
>> 
>> 2) Additionally, some licenses, such as MG-BY-NC (which restricts commercial use of the model), clearly do not comply with the OSD. Should I submit them as complementary materials instead?
>> 
>> 3) Lastly, while I attempted to align MG-BY-OS (where "OS" stands for "Open Source") with OSI’s Open AI definition by requiring that complementary code be licensed under an OSI-approved license, our lawyers opposed this due to concerns about terms incompatibility. Currently, the MG-BY-OS license only mandates open-sourcing the model weights, related code, and scripts (excluding data). I would appreciate any advice on this point.
>> 
>> I participated in the drafting and amendment of ModelGo Licenses and am very willing to discuss any details about them. Please feel free to reach out for discussion. Thanks, everyone!
>> 
>> 
>> Text version of MG-BY at https://ids.nus.edu.sg/modelgo-mg-by.html
>> 
>> All licenses at:
>> NUS ModelGo Licenses Website: https://ids.nus.edu.sg/modelgo.html
>> ModelGo Licenses and FAQ (published by me more info here): https://www.modelgo.li/
>> 
>> 
>> 
>> Best,
>> Moming Duan, 
>> Institute of Data Science, NUS
>> 
>> 
>> 
>> _______________________________________________
>> The opinions expressed in this email are those of the sender and not necessarily those of the Open Source Initiative. Official statements by the Open Source Initiative will be sent from an opensource.org email address.
>> 
>> License-discuss mailing list
>> License-discuss at lists.opensource.org <mailto:License-discuss at lists.opensource.org>
>> http://lists.opensource.org/mailman/listinfo/license-discuss_lists.opensource.org
> _______________________________________________
> The opinions expressed in this email are those of the sender and not necessarily those of the Open Source Initiative. Official statements by the Open Source Initiative will be sent from an opensource.org email address.
> 
> License-discuss mailing list
> License-discuss at lists.opensource.org
> http://lists.opensource.org/mailman/listinfo/license-discuss_lists.opensource.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensource.org/pipermail/license-discuss_lists.opensource.org/attachments/20250211/3043f9ac/attachment.htm>