[License-discuss] [SUBMISSION] AI-MIT License 1.0 — permissive license for AI-generated code

Thu Mar 19 17:23:31 UTC 2026

Andy,

Thanks! I did raise that concern while CERN OHL was drafted (although I'm
sure you already knew), and I am happy that the team paid attention to it.
There are really two issues here. One is CREATION OF NORMS. Capitalized
because I want Open Source folks to keep it in mind. Courts look at
informal norms and codify them, and this is a tremendous risk to us. Second
is that our community, in general, greatly over-estimate the power of
licenses and are unaware of the limits of copyright. The worst case of
course being the "ethical" licenses, but they do it every day. They are an
unconscious force for creating inappropriate norms. They are surprised when
companies blithely ignore their overestimations (although this is not the
only reason companies ignore license obligations) and this creates a
conflict between the spirit and letter of the license. And of course they
are liable to be let down in court if they try to enforce their
overestimations.

    Thanks

    Bruce

On Wed, Mar 18, 2026 at 6:46 AM Shuji Sado <shujisado at gmail.com> wrote:

> Hi,
>
> I agree that this mailing list often reflects a US-law framing, and I also
> agree that differences across jurisdictions matter.
>
> Even so, my own view is that, in practical terms, the treatment of
> AI-generated code may not differ all that dramatically across many major
> jurisdictions, including the US, the EU, and Japan. The doctrinal routes
> may differ, but the basic result is often similar: copyright, if any,
> attaches to the human creative contribution.
> The UK may be a special case because of CDPA 9(3), and China may prove
> more willing to recognize copyright in AI-generated works.
> But I do not think most jurisdictions are moving toward recognizing
> copyright in fully autonomous AI-generated code as such.
>
> If that is broadly right, then the practical consequence is not that
> copyright licensing suddenly stops working. Even if a software tool were
> mostly AI-generated, the license would still operate with respect to the
> remaining human-authored and copyrightable portions. The main difference
> across jurisdictions would be the proportion of the work to which copyright
> attaches, rather than a fundamental change in the overall licensing
> structure.
>
> That is why I am cautious about proposals like AI-MIT/AIAL.
> A license can govern whatever rights exist, but it should not try to
> predetermine, in the license text itself, the legal status of AI-generated
> portions under different national copyright laws, and then attach special
> notice obligations or other conditions on top of that.
>
> That, to me, is where the real risk lies.
>
> Best,
> Shuji
>
> 2026/3/18 20:38 Andrew Katz <andrewjskatz at gmail.com>:
>
>> Hi Bruce
>>
>>
>> On 17 Mar 2026, at 19:38, Bruce Perens via License-discuss <
>> license-discuss at lists.opensource.org> wrote:
>>
>> This is why I started deprecating Open Hardware licenses. Schematics are
>> not currently copyrightable, and what if they were? Perhaps Horowitz and
>> Hill's "Art of Electronics" textbook would be collecting royalties from
>> every electronic product worldwide.
>>
>> Bruce Perens K6BP
>>
>>
>> You raise a very good point, and I have a few observations:
>>
>> 1. (Not surprisingly) this list has a bias towards US copyright law, but
>> it’s worth pointing out that the scope of copyrightability (and similar IP
>> rights) varies significantly from jurisdiction to jurisdiction. In the UK,
>> for example, computer-generated works are explicitly given copyright
>> protection (s9(3) CDPA 1988 as amended [1]). We also have database right
>> (as they do in the EU). Since open source licences are used worldwide, it’s
>> sensible to take a global view where possible.
>>
>> 2. So far as schematics are concerned, they could *potentially*, if
>> complex enough, be regarded as a database of the nodes (although I admit
>> it’s difficult to see how this does square with some of the requirements
>> for database right protection in the UK and EU), but, if hand-drawn, they
>> would almost certainly under UK law, be regarded as an artistic work, and
>> they could, therefore, if machine generated, be regarded as a computer
>> generated work. (Of course, it may be possible to remove the expressive
>> element from the schematics, but there’s currently quite a lot of tension
>> in UK copyright law around the idea/expression distinction. This partially
>> arises from a difficulty in reconciling s9(3) and EU copyright law which
>> requires that the work is the “author’s own intellectual creation” [2]. EU
>> Copyright law still underpins UK copyright law, despite the fiasco of
>> Brexit.
>>
>> This was a major concern when we drafted the CERN OHL because we didn’t
>> want the licence to be seen as trying to exert any control over the use of
>> non-copyright (or other wise non-IP-affected) materials. As it happens, I
>> think we could be more explicit about this point in the FAQs and rationale
>> applicable to the CERN-OHL, and I may propose some amendments accordingly.
>> (On re-reading, I spotted one assumption we made about enforcement of GPL
>> by copyright holders only,  which may not be universally correct any more,
>> so that needs addressing).
>>
>> 3. I’m deeply concerned (and I think this echoes your concerns) that
>> acting as if IP rights exist, when they don’t, can be dangerous. After a
>> while, the courts (and even legislators) can rely on this “custom and
>> practice” to justify the creation of intellectual property rights out of
>> thin air, or to codify rights which never existed in the first place.
>> (We’ve seen this happen to TV formats, for example). I have also seen some
>> attempts by people and organisations to establish a copyright norm which
>> would beneficial to them by publishing papers making certain assumptions. I
>> therefore think that we should be extremely wary of setting licensing norms
>> around AI-generated code, when the underlying ip rights themselves lack
>> clarity.
>>
>> Best
>>
>>
>> Andrew
>>
>> Andrew Katz
>>
>>
>> [1] https://www.legislation.gov.uk/ukpga/1988/48/section/9
>> [2]
>> https://en.wikipedia.org/wiki/Infopaq_International_A/S_v_Danske_Dagblades_Forening
>>
>>
>> _______________________________________________
>> 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
>>
>
>
> --
> Shuji Sado
> Chairman, Open Source Group Japan
> https://opensource.jp/
> English blog: https://shujisado.org/
> Japanese blog: https://shujisado.com/
>
> _______________________________________________
> 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
>

-- 
Bruce Perens K6BP
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensource.org/pipermail/license-discuss_lists.opensource.org/attachments/20260319/8b45c9fe/attachment.htm>