Short Description: A GitHub repo initiated from a Hugging Face Discord discussion on ethical duty-of-care open-source licenses. It explores reasons for their use, notable models under such licenses, and potential integration into Hugging Face's license options.
Extended Description (LLM Summarized and Human edited): This GitHub repository stems from a detailed discussion on the Hugging Face Discord server concerning the implementation and implications of ethical duty-of-care open-source licenses within the Hugging Face ecosystem. The conversation began with community members debating the potential benefits and drawbacks of adopting these licenses, in context of real-world incidents such as the use of AI systems in illegal military occupations against the wishes of the AI systems' original developers. Key points of contention include the clarity and enforceability of ethical licenses, the impact on the open-source community, and the necessity for such licenses in ensuring responsible AI development. Additionally, motivations behind the development of ethical duty-of-care terms in open source licenses were discussed, in the context of difficulties in enforcement of international laws, the comparatively effective enforceability of intellectual property laws, the legacy of ethical duty-of-care licenses in technology (such as Corporate Accountability Lab and HL3,) as well as the constantly evolving intersection of AI with intellectual property law.
Contributors like @Vipidis and @meat have outlined objections, highlighting the vagueness and potential legal challenges of current ethical licenses, while others such as @CalebSol and @TeaDaniels suggest improvements or clarifications to make these licenses more actionable. The discourse also covers possible misunderstandings about ethical licensing, aiming to clear up common misconceptions that might prevent their broader adoption.
The discussion is driving efforts to potentially add ethical duty-of-care licenses to Hugging Face’s default license dropdown when models are submitted. By encouraging these licenses, the community seeks to promote a culture of ethical AI development, fostering accountability and aligning with responsible open-source values. The repository serves as a platform for ongoing dialogue, documentation, and refinement of these ideas to enhance the Hugging Face ecosystem and the broader ML/AI community.
The discussion has evolved out of the Hugging Face Discord, addressing the subject of ethical duty-of-care open source licenses and the potential for adding some of them by name to the default license options dropdown when models are submitted to Hugging Face. Also discussed are reasons for deploying ethical duty-of-care licenses in the Hugging Face ecosystem, prominent models released under ethical duty-of-care licenses (ie Llama 3,) and future directions for developing ethical duty-of-care licenses which benefit the Hugging Face ecosystem and larger ML/AI community.
Please join the discussion here and on the Hugging Face Discord.
Please note this is not officially associated with Hugging Face, and no statements herein are the perspective of Hugging Face, but rather are the expressions of multiple community members who use and respect the Hugging Face ecosystem.