The Hubzilla project has adopted a policy of not allowing contributions based on LLMs or other generative "AI" systems. The policy is loosely based on, and adapted from existing policies by other projects. The full policy reads:
Policy for generative "AI" use in the Hubzilla project
The Hubzilla project prioritizes the social and human aspects of collective software creation. Therefore contributions to the Hubzilla project must not include content generated, in part or in full, by large language models, diffusion models, or similar deep-learning systems. Content, in this context, includes but is not limited to source code, text, and images in Hubzilla Git repositories, merge requests, code reviews and comments.
Contributors to the Hubzilla project may use generative AI tools privately to help comprehend, debug, and review the Hubzilla code base and other content, and to do research related to Hubzilla, so long as they do not contribute content generated by such tools to the Hubzilla core or official repositories.
Third-party dependencies are not directly affected by this policy, but where possible we will seek to use those that align with our policy. Dependencies created entirely or predominantly by generative AI tools will not be used in the Hubzilla project.
If you feel you need to use generative AI in order to communicate effectively, for example if English is not your primary language, then we would rather you submit issues and patches in your native language. We can use translation tools, or may actually speak enough of your language to communicate, and this way we can be confident that any misunderstanding will be on our part and not on the part of tools you used.
This policy has been adapted for the Hubzilla project from the Subplot project generative AI policy and the OpenJDK Interim Policy on Generative AI with input from the Hubzilla community.
The policy was adopted into the
Rationale
Hubzilla is a social project, and we value the human interaction between developers, admins, users and other contributors. Offsetting this to a LLM or similar system, would reduce the social aspect of contributing to the project.
Code and content contributed needs to be maintained, often by other people than the original contributor. Accepting "AI" generated content would open for low quality automated contributions that would potentially require significant work to review and maintain.
Contributions generated using LLMs or other "AI" systems, have numerous issues and uncertainty with regards to licensing. This could put the project into liability if we distribute code we're not allowed to distribute under the license of the project.
Modern "AI" systems, i.e LLMs, diffusion models and other transformer based models generally have numerous ethical issues, like severe environmental impact, shady labour practices, worker exploitation in data labeling, model bias etc, that we feel is incompatible with a social project like Hubzilla.
We want Hubzilla to remain a project made for humans by humans.
#hubzilla #policy #noai
Policy for generative "AI" use in the Hubzilla project
The Hubzilla project prioritizes the social and human aspects of collective software creation. Therefore contributions to the Hubzilla project must not include content generated, in part or in full, by large language models, diffusion models, or similar deep-learning systems. Content, in this context, includes but is not limited to source code, text, and images in Hubzilla Git repositories, merge requests, code reviews and comments.
Contributors to the Hubzilla project may use generative AI tools privately to help comprehend, debug, and review the Hubzilla code base and other content, and to do research related to Hubzilla, so long as they do not contribute content generated by such tools to the Hubzilla core or official repositories.
Third-party dependencies are not directly affected by this policy, but where possible we will seek to use those that align with our policy. Dependencies created entirely or predominantly by generative AI tools will not be used in the Hubzilla project.
If you feel you need to use generative AI in order to communicate effectively, for example if English is not your primary language, then we would rather you submit issues and patches in your native language. We can use translation tools, or may actually speak enough of your language to communicate, and this way we can be confident that any misunderstanding will be on our part and not on the part of tools you used.
This policy has been adapted for the Hubzilla project from the Subplot project generative AI policy and the OpenJDK Interim Policy on Generative AI with input from the Hubzilla community.
The policy was adopted into the
dev branch of the Hubzilla core code repository in commit 387ee1eb, and updated in commit f6add003.Rationale
Hubzilla is a social project, and we value the human interaction between developers, admins, users and other contributors. Offsetting this to a LLM or similar system, would reduce the social aspect of contributing to the project.
Code and content contributed needs to be maintained, often by other people than the original contributor. Accepting "AI" generated content would open for low quality automated contributions that would potentially require significant work to review and maintain.
Contributions generated using LLMs or other "AI" systems, have numerous issues and uncertainty with regards to licensing. This could put the project into liability if we distribute code we're not allowed to distribute under the license of the project.
Modern "AI" systems, i.e LLMs, diffusion models and other transformer based models generally have numerous ethical issues, like severe environmental impact, shady labour practices, worker exploitation in data labeling, model bias etc, that we feel is incompatible with a social project like Hubzilla.
We want Hubzilla to remain a project made for humans by humans.
#hubzilla #policy #noai