<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Why don't agents fit into the description "derivative model
      developed by transferring patterns of weights, parameters,
      activations and/or Output from the Model, such as through
      distillation methods or synthetic data generation techniques, in
      order to replicate, approximate, or otherwise achieve functional
      behavior that is similar to the Model"?</p>
    <p>Is it because you don't consider them models? </p>
    <p>Is it because they don't transfer patterns of weights, etc. or
      create distillations or synthetic data? </p>
    <p>Is it because they don't achieve functional behavior that is
      similar to a model?</p>
    <p>Some other reason?</p>
    <p>One thing to worry about is that the vocabulary for AI is fluid
      and constantly changing. You might not consider an "agent" a
      "model" at the moment, but might it take on more model-like
      characteristics in the future?</p>
    <p>Pam</p>
    <div class="moz-signature">Pamela S. Chestek<br>
      Chestek Legal<br>
      4641 Post St.<br>
      Unit 4316<br>
      El Dorado Hills, CA 95762<br>
      +1 919-800-8033<br>
      pamela@chesteklegal<br>
      <a class="moz-txt-link-abbreviated" href="http://www.chesteklegal.com">www.chesteklegal.com</a><br>
      <br>
      <br>
    </div>
    <div class="moz-cite-prefix">On 12/5/2025 9:17 PM, Moming Duan
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:1B2ED380-0E90-4969-B9D6-B267F105BD2A@gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      Hi Pam,
      <div><br>
      </div>
      <div><br>
      </div>
      <div>I would like to add that, according to the current
        definition, Derivative Materials do not include AI agents
        created using the model and its output. From my personal
        understanding and background, this definition does not present
        significant ambiguity, but I am not certain whether the scope
        might be interpreted more broadly from a legal perspective. If
        this Output-related definition introduces further ambiguity, I
        will consider removing it.</div>
      <div><br>
      </div>
      <div><br>
      </div>
      <div>Best,</div>
      <div>Moming</div>
      <div>
        <div>
          <div><br>
            <blockquote type="cite">
              <div>On Dec 6, 2025, at 10:23, Moming Duan
                <a class="moz-txt-link-rfc2396E" href="mailto:duanmoming@gmail.com">&lt;duanmoming@gmail.com&gt;</a> wrote:</div>
              <br class="Apple-interchange-newline">
              <div>
                <meta http-equiv="content-type"
                  content="text/html; charset=UTF-8">
                <div
style="overflow-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;">Hi
                  Pam,
                  <div><br>
                  </div>
                  <div><br>
                  </div>
                  <div>For clarification, the definition of Derivative
                    Materials does not include the Output of the model
                    (at least that is my intention). It includes
                    derivative models created by transferring patterns
                    from the Output. However, I am reconsidering this
                    definition: as more AI agent systems are built on
                    top of model outputs, I am concerned that including
                    \u201cderivative models developed by transferring
                    patterns of Output\u201d may unintentionally broaden the
                    scope of what counts as a derivative, thereby
                    restricting downstream innovation and conflicting
                    with the Open Source spirit, which is not the
                    intention of this license. </div>
                  <div><br>
                  </div>
                  <div>I would appreciate further discussion on this
                    issue, including the potential pros and cons of
                    removing this Output-related element from the
                    current definition. Thanks.</div>
                  <div><br>
                  </div>
                  <div><br>
                  </div>
                  <div>Best,</div>
                  <div>Moming<br id="lineBreakAtBeginningOfMessage">
                    <div><br>
                      <blockquote type="cite">
                        <div>On Dec 6, 2025, at 01:22, Pamela Chestek
                          <a class="moz-txt-link-rfc2396E" href="mailto:pamela@chesteklegal.com">&lt;pamela@chesteklegal.com&gt;</a> wrote:</div>
                        <br class="Apple-interchange-newline">
                        <div>
                          <meta http-equiv="Content-Type"
                            content="text/html; charset=UTF-8">
                          <div> I renew my objections to this license
                            and the Attribution-ShareAlike version for
                            the same reasons I objected to the previous
                            version, starting with this email,
                            <a class="moz-txt-link-freetext"
href="https://lists.opensource.org/pipermail/license-review_lists.opensource.org/2025-May/005766.html"
                              moz-do-not-send="true">https://lists.opensource.org/pipermail/license-review_lists.opensource.org/2025-May/005766.html</a>,
                            and as explained further in later emails in
                            the thread. I do not believe that putting
                            conditions on output of a model is workable
                            and, where the output is not a derivative
                            work under copyright law, it violates OSD9,
                            "License Must Not Restrict Other Software."
                            <br>
                            <br>
                            The change "Narrows the definition of
                            'Derivative Materials' by including the
                            phrase: 'in order to replicate, approximate,
                            or otherwise achieve functional behavior
                            that is similar to the Model'" does not
                            address this problem and, in fact,
                            exacerbates it. Output that will "replicate,
                            approximate, or otherwise achieve functional
                            behavior that is similar to the Model"
                            identifies output that is highly likely to
                            not be a derivative work under any stretch
                            of the imagination and therefore is well
                            beyond the acceptable reach for an open
                            source license.<br>
                            <br>
                            I do not believe these two versions of the
                            license can be approved.<br>
                            <br>
                            Pam<br>
                            <br>
                            <div class="moz-signature">Pamela S. Chestek<br>
                              Chestek Legal<br>
                              4641 Post St.<br>
                              Unit 4316<br>
                              El Dorado Hills, CA 95762<br>
                              +1 919-800-8033<br>
                              <a
class="moz-txt-link-abbreviated moz-txt-link-freetext"
                                href="mailto:pamela@chesteklegal.com"
                                moz-do-not-send="true">pamela@chesteklegal.com</a><br>
                              <a class="moz-txt-link-abbreviated"
                                href="http://www.chesteklegal.com/"
                                moz-do-not-send="true">www.chesteklegal.com</a><br>
                              <br>
                              <a
href="https://calendly.com/pamela-chesteklegal/30min"
                                moz-do-not-send="true">Set a meeting
                                with me</a></div>
                            <div class="moz-cite-prefix">On 12/5/2025
                              9:04 AM, McCoy Smith wrote:<br>
                            </div>
                            <blockquote type="cite"
cite="mid:e325347c-0c66-4294-bfe9-0d01dddbd58f@lexpan.law">
                              <meta http-equiv="Content-Type"
                                content="text/html; charset=UTF-8">
                              <p>I'm going repeat my comments on the
                                MG-0 license here since they are equally
                                applicable to this license (which
                                appears to replicate the text of MG-0,
                                except for the addition of the
                                conditions in 2.2):</p>
                              <p>1. The disclaimers are not made
                                "conspicuous" as that term is defined in
                                UCC 2-316: <a
                                  class="moz-txt-link-freetext"
href="https://www.law.cornell.edu/ucc/2/2-316" moz-do-not-send="true">https://www.law.cornell.edu/ucc/2/2-316</a>
                                That has been interpreted as requiring
                                something like ALL CAPS or bold, or a
                                different color, or a box (although the
                                criteria changed in 2022). This isn't
                                necessarily a flaw (whether UCC is
                                relevant to open source licenses is an
                                interesting question) but the practice
                                seems to be that most newer open source
                                licenses try to adhere to this
                                requirement (most by using ALL CAPS
                                since that tends to be the only way to
                                do this with .txt files or ASCII --
                                which non-lawyers tend to dislike
                                because they interpret it as screaming
                                without understanding why it's done that
                                way).<br>
                                <br>
                                2. I find the way the grants are
                                structured sub-optimal in the way that
                                it handles the right of performance
                                under copyright law. Rather than being
                                in the grant, it is subsumed into the
                                definition of "Distribution/Distribute"
                                and then grants a right to Distribute.
                                All rights are granted (which is good,
                                that way you don't have to rely on
                                implied grants) but you do need to dig
                                into the definitions to get there.</p>
                              <p>As to the Attribution version of the
                                license, my only comment is this license
                                requires in Section 2.2(i) that a copy
                                of the license be provided. This is a
                                fairly common provision of many so
                                called "permissive" or non-copyleft
                                licenses although I've always wondered
                                what value this requirement provides,
                                given that this license is intended (I
                                believe) to be non-copyleft.</p>
                              <p>Otherwise, this license seems OK.</p>
                              <p>McCoy</p>
                              <p>[in my personal capacity and not as a
                                member of the board]</p>
                              <div class="moz-cite-prefix">On 6/18/2025
                                2:31 AM, Moming Duan wrote:<br>
                              </div>
                              <blockquote type="cite"
cite="mid:883778D3-40CF-4DCF-B65F-AD07D9F427AD@gmail.com">
                                <meta http-equiv="content-type"
                                  content="text/html; charset=UTF-8">
                                Dear OSI Community,
                                <div><br>
                                </div>
                                <div><br>
                                </div>
                                <div><span
                                    style="caret-color: rgb(0, 0, 0);">Following
                                    our previous discussions in May, I
                                    have made further revisions to the
                                    ModelGo </span>Attribution<span
                                    style="caret-color: rgb(0, 0, 0);"> License
                                    (MG-BY-2.0). I am submitting this
                                    updated version for OSI review via
                                    this email. The license text is
                                    attached.</span></div>
                                <div><br>
                                </div>
                                <div>
                                  <div
                                    style="caret-color: rgb(0, 0, 0);"><font
                                      color="#ff0000">\u2014\u2014\u2014\u2014\u2014\u2014 Major
                                      Updates to Previous Submission</font></div>
                                  <div
                                    style="caret-color: rgb(0, 0, 0);"><font
                                      color="#ff0000"><br>
                                    </font></div>
                                  <div
                                    style="caret-color: rgb(0, 0, 0);">
                                    <li><font color="#ff0000">Removes
                                        restrictions on model output.</font></li>
                                    <li><font color="#ff0000">Revises
                                        the termination clause to
                                        provide for automatic
                                        termination.</font></li>
                                    <li><font color="#ff0000">Adds more
                                        explicit granting of rights in
                                        Section 2.1. </font></li>
                                    <li><font color="#ff0000">Narrows
                                        the definition of \u201cDerivative
                                        Materials\u201d by including the
                                        phrase: \u201cin order to replicate,
                                        approximate, or otherwise
                                        achieve functional behavior that
                                        is similar to the Model.\u201d </font></li>
                                    <li><font color="#ff0000">Removes
                                        \u201cDerivative Materials\u201d in
                                        Section 5: \u201cNothing in this
                                        License permits You to modify
                                        this License as applied to the
                                        Licensed Materials.\u201d </font></li>
                                    <li><font color="#ff0000">Fixes
                                        typos and formatting issues.</font></li>
                                  </div>
                                </div>
                                <div><br>
                                </div>
                                <div><span
                                    style="caret-color: rgb(0, 0, 0);">\u2014\u2014\u2014\u2014\u2014\u2014 </span><span
                                    style="caret-color: rgb(0, 0, 0);">License </span><span
                                    style="caret-color: rgb(0, 0, 0);">Introduction</span></div>
                                <div><b><br>
                                  </b></div>
                                <div><b>License Name</b>:<span
                                  class="Apple-tab-span"
                                  style="white-space: pre;">               </span><span
                                    style="caret-color: rgb(0, 0, 0);">ModelGo </span>Attribution<span
                                    style="caret-color: rgb(0, 0, 0);"> License</span></div>
                                <div><span
                                    style="caret-color: rgb(0, 0, 0);"><b>Version</b>: <span
                                    class="Apple-tab-span"
                                    style="white-space: pre;">                             </span>2.0</span></div>
                                <div><font><b>Short Identifier: <span
                                      class="Apple-tab-span"
                                      style="white-space: pre;">           </span></b>MG-BY-2.0</font></div>
                                <div><b
                                    style="caret-color: rgb(0, 0, 0);">Copyleft:</b><span
                                  class="Apple-tab-span"
style="caret-color: rgb(0, 0, 0); font-weight: bold; white-space: pre;">                   </span><span
                                    style="caret-color: rgb(0, 0, 0);">No</span></div>
                                <div><b>Legacy or New</b>: <span
                                  class="Apple-tab-span"
                                  style="white-space: pre;">               </span>New
                                  License</div>
                                <div><b>Drafted By Lawyer</b>: <span
                                  class="Apple-tab-span"
                                  style="white-space: pre;">       </span>Yes, Rajah
                                  &amp; Tann Singapore LLP</div>
                                <div><b>Approved or <span
                                      style="caret-color: rgb(0, 0, 0);">Used</span> by
                                    Projects</b>: <span
                                  class="Apple-tab-span"
                                  style="white-space: pre;">       </span>No</div>
                                <div><br>
                                </div>
                                <div><b>License URL</b>:<span
                                  class="Apple-tab-span"
                                  style="white-space: pre;">                               </span><a
href="https://ids.nus.edu.sg/modelgo-mg-by.html" moz-do-not-send="true"
                                    class="moz-txt-link-freetext">https://ids.nus.edu.sg/modelgo-mg-by.html</a></div>
                                <div><b>Introduction and Video</b>:<span
                                  class="Apple-tab-span"
                                  style="white-space: pre;">       </span><a
                                    href="https://www.modelgo.li/"
                                    moz-do-not-send="true"
                                    class="moz-txt-link-freetext">https://www.modelgo.li/</a></div>
                                <div><br>
                                </div>
                                <div><b>Overview</b>:</div>
                                <div><br>
                                </div>
                                <div>ModelGo Attribution License Version
                                  2.0 (MG-BY-2.0) is a new license
                                  designed for publishing models
                                  (typically neural networks like
                                  Llama2, DeepSeek). It is one of the
                                  variants in the ModelGo License
                                  family. MG-BY-2.0 is the a permissive
                                  license in the ModelGo family,
                                  requiring that the original license <font
                                    color="#ff0000">and attribution</font> be
                                  provided when distributing the
                                  original Licensed Materials or
                                  Derivative Materials (<span
                                    style="caret-color: rgb(0, 0, 0);">Licensed
                                    Materials and </span><span
                                    style="caret-color: rgb(0, 0, 0);">Derivative
                                    Materials are</span><span
                                    style="caret-color: rgb(0, 0, 0);"> </span>defined
                                  in Clause 1). <font color="#ff0000">A
                                    statement of modification is
                                    required, if applicable.</font></div>
                                <div><font color="#ff0000">(Red content
                                    represents the differences from
                                    MG0-2.0 license)</font></div>
                                <div><br>
                                </div>
                                <div><b>Complies with OSD:</b></div>
                                <div><b><br>
                                  </b></div>
                                <div>OSD 3 Derived Works \u2014 MG-BY-2.0 <span
                                    style="caret-color: rgb(0, 0, 0);"> </span><span
                                    style="caret-color: rgb(0, 0, 0);">Clause
                                    2.1 (a) grants copyright and patent
                                    rights to create derivatives.</span></div>
                                <div>OSD 5 and OSD 6 \u2014 No discrimination
                                  clause is included in MG-BY-2.0.</div>
                                <div>OSD 9 License Must Not Restrict
                                  Other Software \u2014 No such restriction
                                  is included in MG-BY-2.0.</div>
                                <div><br>
                                </div>
                                <div><b>The Gap to Fill:</b></div>
                                <div>Model sharing is very common on the
                                  web, with over 1.4 million models
                                  currently listed on Hugging Face (<a
                                    href="https://huggingface.co/models"
                                    moz-do-not-send="true"
                                    class="moz-txt-link-freetext">https://huggingface.co/models</a>).
                                  However, most of these models are not
                                  properly licensed. When publishing
                                  their models, developers typically
                                  choose from three main options (as
                                  seen in the model license tags on the
                                  Hugging Face website):</div>
                                <div><br>
                                </div>
                                <div>
                                  <ul class="MailOutline">
                                    <li>OSS licenses, e.g., Apache-2.0,
                                      MIT</li>
                                    <li>Open responsible AI licenses
                                      (OpenRAILs),
                                      e.g., CreativeML-OpenRAIL-M, OpenRAIL++</li>
                                    <li>Proprietary Licenses, e.g.,
                                      Llama2, Llama3</li>
                                  </ul>
                                </div>
                                <div><br>
                                </div>
                                <div>However, not all licenses are
                                  well-suited for model publishing.</div>
                                <div><br>
                                </div>
                                <div><b>Why not use OSS licenses? </b></div>
                                <div>Traditional OSS licenses lack clear
                                  definitions regarding machine learning
                                  concepts, such as Models, Output, and
                                  Derivatives created through knowledge
                                  transfer. This ambiguity can result in
                                  certain ML activities (e.g.,
                                  Distillation, Mix-of-Expert) being
                                  beyond the control of the model owner.</div>
                                <div><br>
                                </div>
                                <div><b>Why not use OpenRAILs? </b></div>
                                <div>Recently, Responsible AI Licenses (<a
                                    href="https://www.licenses.ai/"
                                    moz-do-not-send="true"
                                    class="moz-txt-link-freetext">https://www.licenses.ai/</a>)
                                  have been widely advocated to govern
                                  AI technologies, aiming to restrict
                                  unlawful and unethical uses of models.
                                  While I acknowledge the growing need
                                  for such governance, these
                                  copyleft-style restrictions do not
                                  comply with the OSD and may cause
                                  incompatibility with licenses like
                                  GPL-3.0. Another concern is that these
                                  behavioral restrictions may
                                  proliferate within the AI model
                                  ecosystem, increasing the risk of
                                  license breaches.</div>
                                <div><br>
                                </div>
                                <div><b
                                    style="caret-color: rgb(0, 0, 0);">Why
                                    not use Llama2 or Llama3 Licenses?</b></div>
                                <div><font>These licenses are
                                    proprietary licenses that are not
                                    reusable. </font>Furthermore, they
                                  include exclusive terms such as "You
                                  will not use the Llama Materials or
                                  any output or results of the Llama
                                  Materials to improve any other large
                                  language model" and copyleft-style
                                  behavioral restrictions.</div>
                                <div><br>
                                </div>
                                <div>In fact, the dilemma in current
                                  model publishing is the lack of a
                                  general-purpose license for model
                                  developers. Additionally, since no
                                  single license meets diverse model
                                  publishing needs, some developers
                                  resort to using CC licenses with
                                  different elements. However, CC
                                  licenses are ill-suited for this
                                  purpose as they do not grant patent
                                  rights. This motivated the drafting of
                                  ModelGo License family, which provides
                                  different licensing elements similar
                                  to CC but specifically designed for
                                  model publishing.</div>
                                <div><br>
                                </div>
                                <div><b>Comparison with Existing
                                    OSI-Approved Licenses:</b></div>
                                <div>Since I could not find an
                                  OSI-approved model license, I can only
                                  compare MG-BY-2.0 with one similar OSS
                                  license \u2014 Apache-2.0</div>
                                <div>
                                  <div><br>
                                  </div>
                                  <div>
                                    <li
                                      style="caret-color: rgb(0, 0, 0);">MG-BY-2.0
                                      defines licensed materials and
                                      derivative works differently from
                                      Apache-2.0, tailoring them to
                                      models.</li>
                                    <li
                                      style="caret-color: rgb(0, 0, 0);">MG-BY-2.0
                                      can govern the remote access
                                      (e.g., chatbot) scenario.</li>
                                  </div>
                                </div>
                                <div><br>
                                </div>
                                <div>If further comparisons or
                                  supporting evidence are needed to
                                  strengthen my claims, please let me
                                  know. I am more than willing to engage
                                  in further discussions with the OSI
                                  community about this license and
                                  contribute to promoting standardized
                                  model publishing. <span
                                    style="caret-color: rgb(0, 0, 0);">\U0001f917</span></div>
                                <div><br>
                                </div>
                                <div><br>
                                </div>
                                <div>Best,</div>
                                <div>Moming</div>
                                <div><br>
                                </div>
                                <br>
                                <fieldset
                                  class="moz-mime-attachment-header"></fieldset>
                                <meta http-equiv="content-type"
                                  content="text/html; charset=UTF-8">
                                <br>
                                <fieldset
                                  class="moz-mime-attachment-header"></fieldset>
                                <pre class="moz-quote-pre" wrap="">_______________________________________________
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
<a class="moz-txt-link-abbreviated moz-txt-link-freetext"
href="mailto:License-review@lists.opensource.org" moz-do-not-send="true">License-review@lists.opensource.org</a>
<a class="moz-txt-link-freetext"
href="http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org"
                                moz-do-not-send="true">http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org</a>
</pre>
                              </blockquote>
                              <br>
                              <fieldset
                                class="moz-mime-attachment-header"></fieldset>
                              <pre class="moz-quote-pre" wrap="">_______________________________________________
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
<a class="moz-txt-link-abbreviated moz-txt-link-freetext"
href="mailto:License-review@lists.opensource.org" moz-do-not-send="true">License-review@lists.opensource.org</a>
<a class="moz-txt-link-freetext"
href="http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org"
                              moz-do-not-send="true">http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org</a>
</pre>
                            </blockquote>
                            <br>
                          </div>
_______________________________________________<br>
                          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.<br>
                          <br>
                          License-review mailing list<br>
                          <a class="moz-txt-link-abbreviated" href="mailto:License-review@lists.opensource.org">License-review@lists.opensource.org</a><br>
<a class="moz-txt-link-freetext" href="http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org">http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org</a><br>
                        </div>
                      </blockquote>
                    </div>
                    <br>
                  </div>
                </div>
              </div>
            </blockquote>
          </div>
          <br>
        </div>
      </div>
    </blockquote>
  </body>
</html>