As expected, President Donald Trump’s administration recently unveiled Winning the Race: AI Action Plan went all in on liberating AI companies to do what they want to make sure “that the United States and its allies win the [AI] race.”
Also: Trump’s AI plan pushes AI upskilling instead of worker protections – and 4 other key takeaways
What about safety and responsibility? Not so much.
The contradiction…
The AI Action Plan states it is “a national security imperative for the United States to achieve and maintain unquestioned and unchallenged global technological dominance,” and then it strongly encourages that “open-source and open-weight AI models are made freely available by developers for anyone in the world to download and modify.”
Also: How the Trump administration changed AI: A timeline
Am I the only one seeing a contradiction with “unchallenged technological dominance” and open-source code and models to be made “freely available” for anyone in the world?
Of course, open-source software’s track record speaks for itself. According to the January 2024 Harvard Business School study, if companies had to create equivalent technologies, rebuilding open-source software would cost $8.8 trillion. Our servers, our clouds, and pretty much our entire IT infrastructure now run on Linux and open-source software.
Still, open source is a vital catalyst for innovation
Trump’s AI Action Plan frames open-source and open-weight AI models as vital catalysts for innovation. This is true. While we can, and have, argued until we’re blue in the face about what the proper open-source AI definition is, AI is built solidly on the framework of open-source programs such as TensorFlow and PyTorch.
As Chris Wolfe, Broadcom’s global head of AI and advanced services, recently said, “Thanks to open source, the pace of AI innovation now surpasses anything closed models could have achieved.”
Also: I found 5 AI content detectors that can correctly identify AI text 100% of the time
The administration also asserts that these freely available models empower not only tech giants, but also small startups, academic researchers, and government agencies. By providing access to the underlying code and model weights, open-source AI is positioned as a cornerstone for advancing the field of AI.
The policy statement argues that open models are not only engines of domestic innovation, but have emerging geostrategic value as global technical standards. This move is meant to counter China’s growing AI influence, especially open models like DeepSeek.
The plan outlines what needs to be done to promote a robust open-source AI ecosystem. This includes increasing access to high-powered computing resources, which were previously locked behind expensive, long-term contracts.
So, who’s going to pay?
That’s easy to say, but who’s going to pay for all that high-priced compute power?
This will be done via the National AI Research Resource (NAIRR) pilot program, which is backed in turn by the National Institute of Standards and Technology (NIST), the National Science Foundation (NSF), and the Office of Science and Technology Policy. These will then partner with tech companies to provide academic and startup communities with advanced model access.
US-based individuals and open-source groups can apply for access to AI resources for start-up projects, request access to educational platforms (such as computational notebooks), and browse curated datasets, pre-trained models, and additional tools for training and testing their AI systems.
Also: AI leaders must take a tight grip on regulatory, geopolitical, and interpersonal concerns
This all sounds impressive, but NAIRR, a public-private project, doesn’t have any specific funding from the recently passed US budget, also known as the Big Beautiful Bill (BBB). Instead, it’s running on the $30 million that former President Joe Biden gave it and the $30 million that Nvidia and other tech companies donated. As a result, only 35 out of more than 150 proposals were awarded NAIRR resources. There’s a lot of strong demand, but nothing like enough funding.
While early drafts of the BBB proposed a $500 million pool of federal funding specifically tied to AI deployment and infrastructure, some of which was to go to NAIRR, in the end, that was dropped. Today, there is no NAIRR funding stream.
Indeed, while there is AI funding in the BBB, it comes with many strings attached. These include compliance with strict domestic content rules and prohibitions on involvement by “prohibited foreign entities.” Thus, as the law firm Ropes & Gray warns, “Companies must invest in due diligence, supply chain transparency, and ongoing monitoring to avoid the risk of losing federal benefits, facing enforcement actions, or losing important partnerships. Technology licensing arrangements, M&A, debt, and restructuring activities must be carefully structured to avoid inadvertent violations. “
So, while the AI Action Plan frees AI companies’ hands to do pretty much whatever they wish, businesses and open-source organizations seeking federal funding will find themselves hampered by a lack of funds and numerous restrictions.
<!–>
‘Founded on American values’
By supporting open models “founded on American values,” the government hopes to set standards that will inform not only domestic, but also global, AI development.
What does that mean? I don’t have a good answer. I know what it doesn’t mean. The AI Action Plan makes it clear that if your plan involves “misinformation, Diversity, Equity, and Inclusion, and climate change,” it won’t be approved.
The plan acknowledges the competitive tension between the proprietary models championed by tech giants and the open-source alternatives. It asserts that open source isn’t just a defensive hedge against domestic monopolies, but a strategic offensive to shape international AI norms as US competitors, such as state-backed Chinese models, push their own open platforms.
Also: Someone used AI to impersonate a secretary of state – how to make sure you’re not next
The plan, however, stops short of mandating open-source releases. Those decisions are left to developers and their organizations. Nevertheless, the federal government will reward and support the open AI ecosystem. How? Through the Department of Commerce’s National Telecommunications and Information Administration (NTIA), the government aims to encourage small and medium-sized businesses (SMBs) to adopt open-source AI software and open-weight AI model. However, there are no particulars yet.
In charge of all this, though, is Marco Rubio, who’s also the Secretary of State, National Security Advisor, and Acting Archivist of the US. It should be noted he has no technical or professional experience in open-source or AI research experience.
What does the AI Action Plan really mean for open-source AI?
I think it can be best summed up as “Good wishes, but almost nothing in the way of concrete support.”
Get the morning’s top stories in your inbox each day with our Tech Today newsletter.
