Following the revocation of the Biden administration’s Executive Order on AI, the Trump administration has released America’s AI Action Plan. The Plan outlines an ambitious vision: “We need to establish American AI—from our advanced semiconductors to our models to our applications—as the gold standard for AI worldwide and ensure our allies are building on American technology.” One proposed path to this goal is through open-source and open-weight AI. While open practices can play a valuable role in accelerating innovation, they are not the only—nor necessarily the most important—means of achieving global leadership.
Open-source innovation represents a form of public innovation. Public and private innovation each play vital and complementary roles in advancing economic growth, competition, and technological leadership. Publicly-funded and open research lays the foundation for breakthrough technologies such as the internet, GPS, and advanced medical treatments by enabling high-risk, long-term exploration. Private firms then build on that foundation, investing in applied R&D to commercialize and scale new technologies. Together, public and private innovation form a dynamic ecosystem in which discovery and deployment reinforce each other and drive sustained progress.
By emphasizing open-source and open-weight AI, the Plan risks overlooking the essential role of intellectual property (IP) protections in supporting private innovation. Although openness can empower startups and small entities to enter the AI marketplace, these organizations also need mechanisms—such as IP rights—to protect their innovations and attract investment. When cutting-edge AI models and tools are released under open-source licenses that permit unrestricted copying and modification, there is no guarantee that the economic benefits will accrue to the original innovators. The rapid and low-cost development of DeepSeek, reportedly aided by Meta’s open-sourcing of its model and access to OpenAI’s distillation features, illustrates this possible risk.
It may be that the Plan uses terms like “open-source” and “open” in a broader sense, where sharing is encouraged under specific conditions, including licensing restrictions or fees, rather than complete and unrestricted free public access. If so, greater clarity would be beneficial. The Plan would be strengthened by explicitly recognizing that IP protection is an essential tool for fostering innovation and ensuring American leadership in AI. At a minimum, the Plan should acknowledge the importance of private innovation, the role of IP rights, and the need to strategically deploy both open and proprietary licensing models. One approach for articulating such licensing options is the Access Methodology Framework, described in my paper “Unlocking AI Transparency: The Access Methodology Framework for Classifying AI System Openness.”
The Plan directs NTIA to “convene stakeholders to help drive adoption of open-source and open-weight AI models by small and medium-sized businesses.” I hope in doing so NTIA will recognize the importance of IP rights incentivizing both public and private innovation and advancing America’s leadership in AI.
