Original title: Moderating Model Marketplaces: Platform Governance Puzzles for AI Intermediaries
Authors: Robert Gorwa, Michael Veale
In the realm of AI, platforms like Hugging Face simplify model access for users, streamlining AI development. But with great convenience comes a great challenge: managing these platforms. This article dives into the tricky task of governing these AI marketplaces. These hubs host countless models and data, potentially leading to misuse or illegal activities. Three platforms—Hugging Face, GitHub, and Civitai—serve as case studies, revealing the governance hurdles faced in moderating models. From licensing to automated moderation, industry practices are explored in response to these challenges. However, there’s still ground to cover. The article proposes avenues for better regulation, suggesting ways platforms can act as responsible gatekeepers. While daunting, these governance puzzles call for platforms to step up, ensuring a balance between accessibility and responsible AI use.
Original article: https://arxiv.org/abs/2311.12573