[License-review] Adapting the license-review process to AI

[Roland Turner]

On 10/9/24 17:46, Stefano Zacchiroli wrote:

> There's another related practical question: what would be the published
> *output* of OSAID-related reviews?
>
> For traditional license review, the output is
> https://opensource.org/licenses , i.e., a (tagged) list of licenses,
> each pointing to the full license text. For OSAID, I'm hence assuming
> (but would like to have confirmation) that it will be a list of "terms
> of use".
>
> One practical problem is that terms of use are less often properly
> versioned than licenses, so we will probably need to both self-host
> (which we already do) and possibly self-version (which would be new).
> Or maybe insist that submitters properly version terms of use as a
> pre-condition for evaluation.

This seems like a minor issue. There might literally be terms of use (as 
in a data use agreement), but as multiple areas of law are involved, 
multiple types of terms in multiple types of instruments may arise. 
OSAID therefore just refers to "an AI system made available under terms 
and in a way that grant the freedoms to". So perhaps "terms of 
availability" but really just "terms". Perhaps hosted at 
https://opensource.org/aiterms .

Anything that OSI approves will presumably be posted on the OSI website. 
The likely standard is the one already used for this list 
<https://opensource.org/licenses/review-process>:

> Provide a *unique name* for the license, preferably including the 
> version number.

I suspect that in pretty short order, someone who gets a version of AI 
system terms with a unique name approved by OSI, and then publishes 
modified terms under the same name and claiming OSI approval will 
receive a polite letter from OSI.

- Roland

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensource.org/pipermail/license-review_lists.opensource.org/attachments/20240910/b2d690cb/attachment-0001.htm>