This proposal creates a National Artificial Intelligence Research Resource (NAIRR) to give U.S. researchers and students fair access to powerful computing, shared data, and tools. It aims to grow America’s AI research capacity and support testing and evaluation of AI systems. The National Science Foundation (NSF) must set it up within one year of enactment.
NAIRR would offer cloud and other computing, open‑source software, an API for AI models, shared data with clear standards, curated datasets, an open data commons, training, user support, and access to AI testbeds for measuring and benchmarking AI systems. NSF will create a Program Management Office and choose a nongovernmental operator to run daily operations, keep a user portal, set policies, publish annual performance reports, and train new users. Eligible users are U.S.-based and tied to a college, nonprofit, federal agency, federally funded research center, or a small business with federal research funding; people working for certain foreign governments are not allowed access. The program must protect privacy and civil rights, audit projects, reserve a significant share of computing for work on safety and trustworthy AI, follow federal scientific‑integrity rules, set cybersecurity requirements using NIST’s framework, and follow federal research‑security policy. It may charge fees but must include a free tier to support research .
- Who is affected: U.S. researchers, educators, and students; certain federal employees and small businesses with federal research awards; other categories may be approved by NSF and the White House science office.
- What changes: Centralized access to compute, data, training, and AI testbeds; an operator runs a user portal and issues public reports; priority for projects on privacy, safety, and trustworthy AI; strong cybersecurity and research‑integrity rules .
- When: NAIRR must be established within one year of enactment; any fees must include a free tier to keep research accessible .
