On July 23rd, the Trump Administration released an “AI Action Plan” with a stated goal of “winning the AI race.” It will serve as a blueprint for policy actions the administration will take to accelerate AI in the United States. While the plan covers a lot of topics unrelated to Open Source, it includes a specific section titled: “Encourage Open Source and Open Weight-AI.” It also recommends federal agencies work with academia and the broader research community to increase AI development, adoption, and evaluation. In this blog post, we explain the areas of the Plan that may impact the Open Source community and highlight recommendations to watch as the administration moves forward with implementation.
Plan Section: Encourage Open Source and Open Weight-AI
This section recommends stakeholder convenings to drive adoption of Open Source and open weight models by small and medium-sized businesses. The effort will be run by the National Telecommunications and Information Administration (NTIA) which ran the public comment period on open weight models under the Biden Administration.
It also recommends expanding access to compute power and other resources (software, data, etc.) for academics and startups through the National AI Research Resource (NAIRR), currently a pilot project at the National Science Foundation with bipartisan congressional support. OSI has publicly supported the NAIRR through our joint public comments with the Open Forum for AI.
Cross-section recommendations where Open Source/foundations could play a role or be implicated
Beyond the specific section on Open Source and Open Weight-AI, the Plan makes several other comments and recommendations the Open Source community should be aware of. Specifically, the Plan:
Notes that the computing stack for AI data centers and infrastructure should be built on American products and free from “foreign adversary information and communications technology and services (ICTS)—including software and relevant hardware.”
Recommends the development of full-stack AI export packages through partnerships between industry consortia and the Department of Commerce.
Recommends the use of regulatory AI sandboxes where AI tools can be deployed and evaluated and the results shared openly.
Recommends the development of AI standards for domain-specific use cases (agriculture, energy, health, etc.).
Makes a number of recommendations to advance how AI models are measured and evaluated, including publishing guidance for federal agencies to conduct evaluations.
Recommends federal agencies work with AI developers to enable the private sector to protect models from security risks.
Recommends the sharing of AI-security threats and vulnerabilities across government and industry through existing and new channels.
Recommends the development of guidance for the private sector on mitigating AI-specific threats and vulnerabilities.
Some of the recommendations above will be implemented by the National Institute of Standards and Technology (NIST). OSI will track public comment opportunities and stakeholder engagement efforts and share them with the Open Source community.
The Plan also includes a number of recommendations related to AI skill development and literacy. It remains to be seen how these recommendations will be implemented with a smaller U.S. Department of Education.
Source: opensource.org