Re: Signed and updated ballot option: Call the Free Software community to develop Free AI.

On Tue, Jul 28, 2026 at 05:46:12AM +0900, Charles Plessy wrote:
>...
> We affirm that:
>
>  * AI models should not be used against the authors of their training data.
>...

That's worse than the previous version, even an author who voluntarily 
provided training data could later take legal action against a competitor
who uses it.

>From the previous version of your GR:

>> * AI models should not be trained on data from authors who clearly  disagree
>>   with this use.

An example:

https://github.com/non-ai-licenses/non-ai-licenses

  4. The source code and the binary form, and any modifications made to 
  them may not be used for the purpose of training or improving machine 
  learning algorithms, including but not limited to artificial 
  intelligence, natural language processing, or data mining. This 
  condition applies to any derivatives, modifications, or updates based 
  on the Software code. Any usage of the source code or the binary form 
  in an AI-training dataset is considered a breach of this License.

It is obvious that this is not DFSG-free.

Do you want to make the "may not be used for the purpose of training"
DFSG-free?

Do even you want to make the "may not be used for the purpose of improving"
DFSG-free?

>  * Can it be interpreted as violating the DFSG ?  Yes, but it also has
>    interpretations that do not violate the DFSGs.  Please assume I
>    mean those.
>...

Voters will vote on the text, not on what the proposer meant.

And an interpretation that does not violate the DFSG would have the 
result that we affirm a right of authors, but when an author writes
that into the license we will remove the package from our distribution.

> Have a nice day,
> 
> Charles

cu
Adrian