[License-discuss] [DISCUSSION] AIAL v2 (was AI-MIT) discussion — permissive license + provenance declarations + limited no-claim framework

[Pamela Chestek]

It seems that you are trying to do three different things with one 
document: (1) create a system for identifying the provenance of code, 
(2) apply a license and (3) optionally state that someone is waiving a 
claim to copyright they might have.

I don't see any reason why purpose (1) is tied to purposes (2) and (3). 
For example, why couldn't someone disclose that code was AI assisted and 
that the license is the Apache license, GPL, MIT, or Unlicense? As to 
(2) and (3), the legal mechanisms are different but how they are 
implemented is the same, by having a written document stating what the 
legal mechanism is. Doing (2), and (3) optionally in the same document, 
is unnecessarily complicating things.

Purpose (1) is laudable and would benefit from an industry-wide 
initiative to define how to identify when AI has been used and to what 
extent. It could conceivably be a practice similar to, or part of, SPDX. 
But the only relevance is whether that code can be subject matter of a 
license. That information doesn't limit the next step, what permissions 
users can have, so in my view it is better for the contributor to have 
the freedom of license choice separate and apart from the provenance 
information.

Further, the provenance provision still relies on the adoption of some 
kind of tagging -- how and where that's implemented isn't clear. 
Moreover, the provenance information isn't mandatory, and even when it 
has been provided it has no legal effect. So really the only legally 
operative language you have is a poor attempt at rewriting the MIT license.

Pam

Pamela S. Chestek
Chestek Legal
4641 Post St.
Unit 4316
El Dorado Hills, CA 95762
+1 919-800-8033
pamela at chesteklegal.com
www.chesteklegal.com


On 3/28/2026 5:28 AM, Nik wrote:
> Hello all,
>
> Thank you to everyone who commented on the earlier AI-MIT / AIAL 
> submission and the subsequent discussion.
>
> I decided to start from scratch in a new thread to make it clear.
> Updated repo with current docs: 
> https://github.com/aicrafted/AI-Attribution-License
> New edition of license text, provenance and faq also attached to letter.
>
> IMO the most important points from the previous thread were:
> 1. The original AI-MIT name was not appropriate and created avoidable 
> confusion.
> 2. A single project-level authorship declaration is not sufficient for 
> real repositories.
> 3. Per-file or per-artifact provenance may be useful as documentation, 
> but it should not be treated as a conclusive legal determination.
> 4. The earlier draft also relied too heavily on hypothetical SPDX 
> evolution.
> 5. The concept needs a cleaner separation between provenance 
> disclosure and legal effect.
>
> Based on that feedback, considering a narrower v2 direction:
>
> - A conservative permissive license core, intentionally close in 
> spirit to MIT/ISC
> - An optional provenance declaration layer, used as documentation and 
> contributor representation
> - An optional contributor-limited no-claim / covenant layer for 
> specifically declared generated-origin contributions
> - An explicit rule that provenance declarations:
>   - do not determine legal status by themselves
>   - do not negate third-party or unknown rights
>   - do not expand permissions beyond the declaring contributor’s own 
> rights
>
> In other words, the revised direction is not a license that decides 
> whether AI-generated code is public domain but rather: a permissive 
> license framework that allows provenance-aware disclosure and, where 
> expressly chosen, a contributor-limited no-claim posture, without 
> pretending to conclusively resolve unsettled authorship law.
>
> At this point, I would like to focus on questions:
> 1. Is it preferable to keep provenance syntax entirely in a separate 
> specification, rather than trying to embed those semantics directly in 
> the license text
> 2. Does the "contributor-limited no-claim / covenant" model seem 
> materially safer than the earlier "fully AI-generated => public 
> domain" framing
> 3. Are there obvious pitfalls in treating `mixed`, `unknown`, and 
> `inherited` as explicit conservative states that do not imply any 
> special legal effect
>
> Thank you again for the comments — they were useful, and the goal here 
> is to narrow the scope and address the real concerns.
>
> Best regards,
> Nik Babichev (Nik the human)
>
> _______________________________________________
> 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