From 7f0dab818b591e5cd9c2de08e24f6363e23ec3f1 Mon Sep 17 00:00:00 2001 From: generic-pers0n Date: Thu, 10 Apr 2025 19:21:24 +0000 Subject: [PATCH] CONTRIBUTING: Add a policy on AI Signed-off-by: generic-pers0n --- CONTRIBUTING.md | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 83067c678..d779a4691 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -19,6 +19,44 @@ references to `main` below with `audacity-3.7-rebase`.** **This notice (in bold) is temporary. Once removed, we will accept contributions as we did before the rebase again.** +## The Use of LLMs and AI in Contributions +**Please read this section IN FULL before continuing with your contribution. +We reserve the right to close or ignore any issue, feature request, or +contribution if you fail to read this section otherwise.** + +As of 2025, within the last few years, advancements in AI have shown us the +true capabilities we can achieve with AI. It is quite amazing, and also quite +scary, how AI has advanced to this state compared to previous states. However, +with these advancements also come their problems. + +A problem fellow open-source projects have experienced, and, at the time of +writing this section, are still experiencing, is the use of AI to generate +_entire reports_. [This article](https://thelibre.news/foss-infrastructure-is-under-attack-by-ai-companies) +gives a good overview of the situation they are facing. It also includes +the larger infrastructure situation many open-source projects are facing too. + +You may feel _very compelled_ to use AI to generate a bug report for you. **DO +NOT DO THIS** please! This especially applies if you are intending to report a +security vulnerability. As has been the case before, and still continues to be +the case, AI is susceptible to hallucinations, meaning they can give false +information in a report. At the very least, if you are going to use AI +inevitably, please double-check the report to ensure everything is completely +accurate. + +A good use of AI, subjectively, is as a companion or an assistant. Using it to +do your job for you is **NOT** a good idea. Therefore, while we prefer you to +**NOT** use AI in any report, if you, for any reason, are going to inevitably +use AI in a report, please _check_ what the generated report contains. + +If we suspect, with justification, that a report is AI-generated, we reserve +the right to, at worst, ignore your report. We do not have the time to spend +weeding through inaccurate AI-generated reports. + +Similarly, **DO NOT USE AI FOR PULL REQUESTS, ESPECIALLY FOR CODE!** Who +actually owns the code is uncertain, to say the least, and that is the reason +why we don't accept AI-generated code. We will close pull-requests that +otherwise violate this rule. + ## Reporting bugs and feature requests Our IRC channel, [`#tenacity` on Libera](https://web.libera.chat/#tenacity)