DLF Response to Generative AI and D: An Open Letter

[Indraj Gandham]

Thank you for reading the letter and preparing this response. It 
is good to hear that the concerns raised are being taken 
seriously. I cannot speak on behalf of the other signatories, but 
nevertheless I would like to make a number of points:

Disclaimer: I am not a lawyer and this is not legal advice.

First of all, it is correct to point out that this risk is not 
unique to LLM-generated contributions. However, it can also be 
argued that a human would not unknowingly copy to the same extent 
as an LLM would, in the sense that for larger works (e.g. over a 
thousand lines of code) it is unlikely that a human would be able 
to recall all of the original material from memory or recreate it 
by accident. Even so, this aspect does warrant further 
consideration as copying can be done knowingly and maliciously by 
prospective contributors. Having a policy against so-called 
“drive-by PRs” may mitigate this to some extent.

Secondly, while responsibility for the inclusion of infringing 
material does lie with the project’s review process, 
responsibility for further distribution of infringing material 
does lie with the users of Phobos. To my knowledge, whether the 
infringement was innocent or not is relevant only when 
determining the severity of the penalty to be imposed (in the 
United States). It does not erode liability. There also exists a 
risk of secondary economic effects; for example, a commercial 
product which includes infringing portions of Phobos may need to 
be withdrawn, directly affecting sales and therefore the 
livelihoods of the software engineers working at that company. It 
is also likely that the company would sustain reputational 
damage, potentially affecting its ability to raise future revenue.

Thirdly, I would like to draw attention to Section 1202(b) of the 
Digital Millennium Copyright Act, reproduced below:

> (b) Removal or Alteration of Copyright Management 
> Information.—No person shall, without the authority of the 
> copyright owner or the law—
> (1) intentionally remove or alter any copyright management 
> information,
> (2) distribute or import for distribution copyright management 
> information knowing that the copyright management information 
> has been removed or altered without authority of the copyright 
> owner or the law, or
> (3) distribute, import for distribution, or publicly perform 
> works, copies of works, or phonorecords, knowing that copyright 
> management information has been removed or altered without 
> authority of the copyright owner or the law,
> knowing, or, with respect to civil remedies under section 1203, 
> having reasonable grounds to know, that it will induce, enable, 
> facilitate, or conceal an infringement of any right under this 
> title.

The response to our letter does not take into account that 
companies distributing LLM products are being sued under this 
provision. In one of the most high-profile cases, the plaintiffs 
alleged that the defendants’ product alters CMI, and that they 
distributed the product knowing that it would alter CMI. The 
defendants attempted to have the allegation dismissed, but this 
was rejected by the court. The matter is stayed pending referral 
to a different court.

I would also like to draw attention to the phrase “reasonable 
grounds to know”. If DLF is aware of the risk that LLMs may alter 
CMI, and still permits the inclusion of large volumes of LLM 
output in Phobos, would DLF be held liable? Could this also 
affect downstream users distributing Phobos in their products? 
These are questions that need to be answered. On that note, it is 
good to hear DLF is considering our suggestion to seek legal 
guidance.

The policy principles and contributor affirmation represent a 
strong start. It is, however, somewhat disappointing to hear that 
our suggestion regarding a moratorium on the direct inclusion of 
AI outputs (specifically in Phobos) has been rejected outright. 
If this is of immediate concern to any of the signatories, you 
may wish to consider other options such as Adam D. Ruppe’s OpenD 
compilers. There is no need to abandon D altogether. I also 
understand that a new fork of D without LLM output may appear at 
some point in the near future.

Finally, I would like to say that I never expected to receive so 
much support, including from some high-profile users of D. The 
letter may not have succeeded in all of its aims, but the 
response provides more clarity around DLF’s position with regard 
to our concerns. This is important because it allows us to make 
informed choices in the future.

To everyone on both sides of the ledger, thank you for taking the 
time to participate.