[License-review] For Approval: OpenMDW License Agreement, versions 1.1 (OpenMDW-1.1)

[Richard Fontana]

BTW I think the concern I'm raising can be anchored to some degree in
the OSD. OSD 9 says:

"9. License Must Not Restrict Other Software

The license must not place restrictions on other software that is
distributed along with the licensed software. For example, the license
must not insist that all other programs distributed on the same medium
must be open source software."

This of course, like its more problematically worded DFSG antecedent,
has to be read with the prevailing community interpretation of the GPL
in mind.

I don't think it's much of a stretch to think of defensive termination
provisions as "restrictions". This is how they were analyzed in the
GPL community historically (for example, one of the reasons Apache-2.0
was thought to be incompatible with GPLv2 was its patent termination
provision, and (with the caveat I haven't digested McCoy's point yet)
the provision I was referring to in GPLv3, which McCoy quoted, was
designed in part to eliminate that source of license incompatibility.

OSD 9 was one of the main bases of concern about SSPL, because SSPLv1
seemed to require that service providers provide source code not only
of their version of the original SSPL-licensed software but also
various sorts of other programs that I would characterize as being
"separate" in an intuitive sense that informed many discussions in the
past about the proper limits of the scope of copyleft.

OpenMDW is *designed* to be applied to a bundle of things that are
conceptually "separate" (even if "related"), as manifested in the term
"Model Materials":
"Model Materials" means the materials provided to
you under this agreement, consisting of: (1) one or more machine learning
models (including architecture and parameters); and (2) all related artifacts
(including associated data, documentation and software) that are provided to
you hereunder."

I haven't looked closely at existing uses of OpenMDW, but we can
suppose it is expected that you'd use OpenMDW for the whole set of
stuff associated with a public model release. This set of things is,
I'd argue, a bunch of *separate* things in the historical folkloric
FOSS sense. Indeed today it is not uncommon for providers of public
models to put the model under one license (sometimes non-FOSS) and the
software distributed along with the model under a different, FOSS
license. Indeed Mike Dolan explained that that complexity is a
motivation for creating OpenMDW to begin with - to have one license
that is suitable for all the stuff that gets released with a model -
weights, model architecture stuff, sample training code, sample
inference code, documentation, and so forth. (BTW - it may be relevant
to note the continuing aggravating practice of AI engineers, or
whoever is actually making these decisions, of splitting up such
materials across differently hosted repositories.)

The point I'm raising is that to have a copyright (and perhaps even
patent) defensive termination provision apply across the whole set of
conceptually distinct things, such that a claim against one item
triggers termination of rights in another item, is possibly an OSD 9
violation, somewhat masked by the fact that the same license is being
used for everything, by design.

Richard

On Thu, Aug 13, 2026 at 1:58 PM Richard Fontana <rfontana at redhat.com> wrote:
>
> On Thu, Aug 13, 2026 at 1:02 PM Michael Dolan
> <mdolan at linuxfoundation.org> wrote:
> >
> > To the OSI License Review Committee:
> >
> >
> > On behalf of The Linux Foundation, the license steward, I submit the OpenMDW License Agreement, version 1.1 ("OpenMDW-1.1"), for review and approval as a new license. The complete license text is included below in plain text (LICENSE.OpenMDW-1.1) and is also published at https://github.com/OpenMDW/OpenMDW (1.1/ folder) and at https://openmdw.ai/license/1-1/.
>
> I am happy to see OpenMDW submitted for OSI approval since there is
> one issue that has concerned me but I honestly don't know how the
> community would generally react to it. It specifically has to do with
> the defensive termination provision.
>
> > Compared with the Apache-2.0 license, both include a defensive litigation termination provision and expressly address patent rights. However, the termination provisions differ in scope.  Apache-2.0's termination is limited to patent claims, whereas OpenMDW-1.1's termination applies to both patent and copyright assertions (the sole substantive change from OpenMDW-1.0 to OpenMDW-1.1). OpenMDW-1.1 also omits Apache-2.0's NOTICE-file and statement-of-changes requirements, conditioning redistribution only on retention of the license text and applicable notices.
>
> and now I understand that this was the one substantive change from
> OpenMDW-1.0 to OpenMDW-1.1:
>
> > feedback from corporate legal teams on Version 1.0; its only substantive change is extending the litigation-termination provision from patent assertions to patent or copyright assertions. I am only submitting
> > version 1.1 for review at this time, although the same review could include version 1.0 if the review committee prefers to include both.
>
> So this is the provision in question:
>
> > If you file, maintain, or voluntarily participate in a lawsuit against any
> > person or entity asserting that the Model Materials directly or indirectly
> > infringe any patent or copyright, then all rights and grants made to you
> > hereunder are terminated, unless that lawsuit was in response to a
> > corresponding lawsuit first brought against you.
>
> This seems to potentially go further than the more familiar kinds of
> defensive termination provisions we see in modern open source
> licenses. First, it applies to copyright as well as patent assertions.
> Apache-2.0, MPL-2.0, and EPL-2.0 have defensive termination provisions
> limited to patent assertions and only terminating patent licenses, if
> I'm reading them correctly. The latter two's predecessors also I
> believe only had patent-litigation-triggered defensive termination.
> The sole "modern" (or maybe any?) open source license I can think of
> that might have defensive termination extend to copyright litigation
> is GPLv3, but this is not explicit in the license text and I'm
> honestly not sure it should be considered a correct reading (I really
> haven't thought about it, at least recently). GPLv3 explicitly says
> that certain patent litigation claims are examples of "further
> restrictions"; it says nothing about copyright in this regard.
>
> As a side note, but maybe relevant, some legacy (post-1998)
> "corporate" open source licenses with patent-litigation-triggered
> defensive termination provision were controversial and widely
> criticized at the time they were proposed because they sometimes
> extended to patent claims unrelated to the software being licensed.
> While the licenses I'm thinking of (e.g. IBM-PL, CPL) were
> OSI-approved, and reflected the zeal within the open source community
> 25-30 years ago over the software patent problem, policy criticisms
> were influential and later open source licenses tended to have the
> more limited approach to patent license termination seen in the
> licenses I mentioned above.
>
> in OpenMDW-1.1, we seem to see a departure from the general spirit of
> more restrained approaches to defensive termination, in a novel (for
> potential OSI-approved licenses) context that I'm concerned may be
> problematic. Not only does the license extend termination to copyright
> litigation, it also broadens the scope of termination by covering
> seemingly unrelated materials. For example, suppose I believe that an
> OpenMDW-1.1-licensed model infringes my copyrights. I sue the model
> licensor, but now my copyright and patent rights to some Python code
> distributed (in some sense) along with the model are terminated.
>
> This issue seems more acute given the current atmosphere in the open
> source community around the use of open source software in training
> models. Many members of the open source community, rightly or wrongly,
> seem to feel strongly that their work is being misappropriated in the
> model training process and that the output of generative AI models
> potentially infringes their copyrights (which is of course possible in
> principle). Suppose an open source developer determines that an
> OpenMDW-1.1 model generates some software output, say, that they
> believe infringes their copyrighted work which, by hypothesis, was
> used in training the model. Should they have their licenses to not
> only the model but also any related code be terminated because they
> decide to pursue an infringement claim against a provider or
> distributor of the model materials? I don't know, but I do know that
> this is unlike any OSI-approved license I can think of, so I think the
> OSI will need to specifically deliberate over this issue.
>
> Richard