NIST launches AI standards ‘Zero Drafts’ pilot project

The US National Institute of Standards and Technology (NIST) has announced a new initiative – the AI Standards ‘Zero Drafts’ Pilot Project – to accelerate the development of high-quality AI standards and widen participation from across the AI community. The project, first introduced at the NIST AI Symposium in September 2024, aims to produce rigorous, stakeholder-informed drafts of AI standards that address critical technical needs while facilitating broad engagement.

From now through summer 2025, NIST will be scoping and releasing initial concept papers on key standardisation topics. Stakeholders are invited to provide input, particularly on topic selection and draft scoping, to help shape these preliminary ‘zero drafts.’ Once refined through public input, these drafts will be submitted to standards developing organisations (SDOS) for further development into voluntary consensus standards.

The pilot responds to two central challenges in AI standardisation: the need for rapid development of reliable standards and the necessity of inclusive, multi-disciplinary contributions. Traditional SDO processes, while essential for rigour, can be time-consuming. NIST’s approach offers a pathway to streamline consensus-building by developing stakeholder-driven drafts that are already grounded in science and aligned with community needs.

How the Process Works

NIST will:

• Propose initial topics and collect feedback on prioritisation and scope
• Release concept papers for selected topics
• Draft and refine preliminary standards based on stakeholder input
• Submit the final zero drafts to SDOs for formal standardisation processes

Initial Topics in Focus

The pilot will focus on several high-priority areas, including:

• Transparency and documentation practices such as model cards and system documentation
• Testing, evaluation, verification, and validation (TEVV) methods for both traditional and generative AI
• AI system design frameworks, including AI stack mapping and reference architectures
• Risk mitigation for synthetic content, including watermarking and provenance tracking

Open invitation for feedback

NIST encourages feedback on the pilot process, topic prioritisation, and concept development. Suggestions may include additional topics, scoping strategies, or relevant organisational insights. Contributions should be sent to ai-standards@nist.gov, where submissions will be made publicly available.

Organisations are also invited to host listening sessions to support broader engagement during all stages of the process. Interested parties may reach out using the same email address.

For more information and updates, please visit NIST’s AI Standards program page.