Hi Folks,

there are a few things I would do, If I were you,

000) talk to Mischa Peters He is trying to get support for writing an
Imap4 OpenBSD style secure privsep
unveil and pledged daemon... perhaps the money you are putting  into
tokens can provide some incremental
support for a dev.  donate early and often...

if you are going down the AI route
01) instruct it  at the outset to include the ISC BSD style Header,
in any file it creates under your instruction
 in the first instance. because the output of that would be under a
BSD License with all the generosity
 and freedom of it.

0.5) if in doubt get a friend who is collaborating with you to do the
same so it is not just one agent one session
doing everything)
1) remember from the outset if using claude as a tool to translate
your thoughts, concepts design,
managerial direction creative instincts, into draft c code,
2) if you are paying claude you are paying for its output,
3) you are responsible for what claude does ...  (that cuts both ways)
>From a legal standpoint don't concede anything... Claude didn't write
C code of its on volution... you
instructed it guided it, guided  in testing.

4) get input from other real  people and put those changes directly in
the code  (no longer just claude  handling
c code directly)
and if you can provide some support for real people as developers have
to eat too
(unfortunately we are seeing a trend where AI pushers are claiming to
give massive opportunities)
 but no one explains who the AI companies are going to sell to if
everyone is laid off / displaced by it )

5)  if you are not getting traction from  the user  community  then as
a  use a competing model to  audit the code,
take feedback form the  competing model and modify the C code (with
its  ISC license /  BSD License)
 with  any changes that make sense. ) i

6) be ware that AIs have no life and sometimes they will over
complicate things just for the craic (fun as we say in ireland)
it is also helpful in consuming loads of tokens for very little
gain...  be prepared to delete and start over...

best of luck,
Tom





On Sat, 14 Nov 2026 at 12:19, Peter N. M. Hansteen <[email protected]> wrote:
>
> On Fri, Nov 13, 2026 at 10:36:51PM -0800, [email protected] wrote:
> > Hello misc@,
> >
> > I'd like some critical eyes on aimap, an IMAP4rev2 (RFC 9051) server
> > for OpenBSD. To say it first: all of its code was generated by
> > Anthropic's Claude, under my direction. I designed it with the model,
> > reviewed it and tested it, but I did not write the C.
> >
> > I know the project's position on LLM-generated code, so to be clear:
> > I am not proposing it for base or for the ports tree. It claims no
> > copyright of its own. The parts adapted from base (log.c, imsgev.c,
> > parse.y and a few others) keep their original notices, and CREDITS
> > says what came from where.
>
> For the archives, it is not just the project's position, it is dangerous
> territory copyright-wise, with open questions that will likely only be
> decided by horribly expensive lawsuits.
>
> I am not in any way someone who decides what does or does not go into
> OpenBSD, but I think it is worth considering the fact that LLM output
> is not copyrightable (see Cory Doctorow's overview with relevant links
> at [1]).
>
> That means that while you can probably claim copyright of and
> assign a license to your input to the LLM (the prompts), you are not
> entitled to the same for the LLM output.
>
> The next question is where the courts land on whether the LLM training
> process consists of a series of copyright violations or some version
> of "fair use". There is a non-zero chance that fact such as lack of
> explicit consent to that particular use will be found to be a violation.
>
> For those reasons, I think your chances of getting this code into the
> tree are minimal.
>
> That said, it is possible it is possible that you learned quite a bit
> about the subject by doing the prompting.
>
> But any attempt to claim copyright on LLM output will be to invite a world of 
> pain.
>
>
> [1] 
> https://doctorow.medium.com/https-pluralistic-net-2026-10-31-burn-berne-unvictored-ffc900dcb9f3
>
> --
> Peter N. M. Hansteen, member of the first RFC 1149 implementation team
> https://nxdomain.no/~peter/blogposts 
> https://nostarch.com/book-of-pf-4th-edition
> "Remember to set the evil bit on all malicious network traffic"
> delilah spamd[29949]: 85.152.224.147: disconnected after 42673 seconds.
>


-- 
Kindest regards,
Tom Smyth.

Reply via email to