Policy or Community?: Supporting Individual Model Creators' Open Model Development in Model Marketplaces

Community norms matter more than formal policies: The study found that individual model creators' sense of responsibility is "deeply shaped by community norms rather than formal policies," suggesting that platforms should invest in fostering responsible AI communities alongside or instead of relying solely on top-down policy enforcement.

Creators use responsible AI tools for self-protection, not ethics: Rather than adopting responsible AI (RAI) tools to prevent harm, creators primarily use them "for self-protection and visibility," indicating a gap between platforms' intended use of these tools and how creators actually deploy them in practice.

Three critical governance gaps exist: The research identified three key regulatory needs that current policies don't adequately address: (1) reducing downstream harms from models, (2) recognizing creators' contributions and originality, and (3) securing model ownership—suggesting platforms need more targeted interventions in these areas.

Model transparency varies widely, complicating governance: Creators work with "partially open models that vary widely in openness and transparency," making it difficult for platforms to enforce consistent responsible AI standards, since the starting point for each creator differs significantly.

One-size-fits-all policies may backfire: The study suggests that "platforms' governance decisions must consider how policy interventions shape the practices and motivations of individual creators," implying that poorly designed policies could inadvertently discourage responsible behavior rather than encourage it.