[License-review] ModelGo Zero License, Version 2.0

Fri Feb 14 18:47:57 UTC 2025

Moming,

The OSI is not likely to approve a license that has this particular use 
restriction. To understand why, I suggest you read this article: 
https://the.webm.ink/just-obey-the-law

Although the OSI has approved licenses with jurisdictional clauses in 
the past, they are disfavored. Stating a specific jurisdiction is not a 
good idea (as you recognize), since neither of the parties may be in 
that jurisdiction, making it inconvenient for both parties. Trying to 
define it as tied to the licensor's place of business is problematic for 
the reasons Carlo described. Experience has shown that defining 
jurisdiction only causes more problems that it solves.

I consider not only Annex A, but any annex, a problem. Your license 
allows a licensor to add unknown content to the license in the form of 
"annexes" ("'License' means the terms and conditions for use, 
reproduction and distribution as set out in [Sections 1 to 8 and *the 
annexes*] of this License"). The OSI will not approve licenses that are 
not self-contained because of the high likelihood that the added content 
will not comply with the OSD or OSAID. So the possibility of undefined 
"annexes" isn't acceptable.

Further, in my opinion this Annex A is not acceptable. As I understand 
it, Annex A is meant to be redundant to what the text of the license 
says. If so, it is superfluous. However, its existence invites others to 
change an X to a check, or vice versa, changing the actual license 
itself to one that is non-free. For example someone could change the X 
to a check for "Use Restrictions (RAI) on Licensed Materials, Derivative 
Materials and Output" and nevertheless claim that their system is open 
source because they used the ModelGo Zero License. To be approved, 
licenses must be immutable. You have described the Annex as informative 
- it's perfectly fine to use as an educational or informational tool 
elsewhere, but it shouldn't be part of the license itself.

I realize some of my comments seem to go against what lawyers who don't 
work in the field believe are good drafting practices. However, our 
standards are time-tested, with 20 years of analysis and review of these 
licenses to understand what makes them good or bad. We have also learned 
that these licenses have incredibly long lifespans, so it's very 
important to get them as right as possible. As Carlo also noted, this is 
the first candidate for an AI license, so we will be exceptionally 
careful in the review. I haven't looked at it carefully myself yet, but 
I expect that I will have more comments on the drafting once I have. I 
personally am rooting for you, I would be very happy to have an 
OSI-approved OSAID license, so I hope we can all collaborate to make 
this an acceptable license.

Pam

Pamela S. Chestek (in my personal capacity)
Chestek Legal
PLEASE NOTE OUR NEW MAILING ADDRESS
4641 Post St.
Unit 4316
El Dorado Hills, CA 95762
+1 919-800-8033
pamela at chesteklegal.com
www.chesteklegal.com

On 2/13/2025 10:50 PM, Moming Duan wrote:
> Hi Eric,
>
>> On 14 Feb 2025, at 10:56 AM, Eric Schultz <eric at wwahammy.com> wrote:
>>
>> To me, that implies that the usage of licensed works are dependent 
>> upon the usage being legal. That's not a requirement that an OSD 
>> compliant license can have; additionally, it's unnecessary, the state 
>> already enforces those rules, you don't need the license holder to do 
>> so as well.
>
> I am not a lawyer, but my intuition is that a license will be 
> ineffective or unenforceable if its terms do not comply with 
> applicable law.
> Regarding OSD compliance, I think 2.3(a)(i) is not a discrimination 
> clause against persons or groups, as every entity can be sued and 
> suspected of breaking the law.
> My lawyer also advised retaining this clause, as we do not intend for 
> the licensor to be liable for the illegal use of licensed materials.
> I failed to convince my lawyer to remove this clause because I cannot 
> identify who would be harmed by it, and its removal may increase 
> potential risks.
>
>
>> PS: While the Open Source AI definition says you don't have to 
>> include the source data to be an "Open Source AI", I would disagree 
>> with that conclusion. But that's my own two cents.
>
> My personal view is that open-source AI systems require open-source 
> datasets, but open-source models do not. I believe the scope of open 
> source should not extend to parts governed by another license or 
> applicable law, as such proliferation could cause inconsistencies and 
> conflicts in license terms.  As an open-source model, it should, at a 
> minimum, keep its parameters and architecture available and should not 
> prohibit any kind of use of its generated output, such as reverse 
> engineering or distillation.
>
>
> Best,
> Moming
>
> _______________________________________________
> The opinions expressed in this email are those of the sender and not necessarily those of the Open Source Initiative. Communication from the Open Source Initiative will be sent from an opensource.org email address.
>
> License-review mailing list
> License-review at lists.opensource.org
> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensource.org/pipermail/license-review_lists.opensource.org/attachments/20250214/b28968fb/attachment.htm>