On June 12th, Hugging Face submitted a response to the U.S. Department of Commerce's NTIA request for information on AI accountability. The company emphasized that documentation and transparency norms are crucial driving accountability, and that tackling the challenges of rapidly evolving AI requires input from a wide range of stakeholders.
Hugging Face stated that its mission is to democratize good machine learning, making systems not only easier to develop and deploy, but also easier to understand and critique. To that end, it has invested in education, documentation, community guidelines, and low-code tools that allow people with all technical backgrounds to analyze datasets and models.
The company made three key recommendations for accountability mechanisms:
- Focus on all stages of ML development, because societal impact depends on choices made throughout the pipeline, not just at deployment. Late-stage assessments risk incentivizing surface-level fixes that ignore deeper issues.
- Combine internal requirements with external access. Good documentation shapes responsible development, but external verification is needed to ensure claims are accurate and to empower stakeholders outside the development chain.
- Invite participation from the broadest possible set of contributors, including developers, researchers, advocacy organizations, policymakers, and journalists. No single entity can fully understand the transformative impact of rapid AI adoption.
Hugging Face argued that prioritizing transparency in both ML artifacts and assessment outcomes is essential to meeting these goals. The company's full response is available in a linked PDF.