In a new insight, Technology and Innovation Policy Analyst Angela Luna reviews the pros and cons of open- and closed- artificial intelligence (AI) models, and explains the delicate balance policymakers must strike in regulating access to open source for innovation.

Key points:

• Open artificial intelligence (AI) models – those that can be downloaded, inspected, and modified – offer greater access and flexibility of use, while lowering costs and barriers to AI innovation compared with closed models; many policymakers, however, favor closed models over open models based on the assumption that they offer greater safety and control.
• As open models – particularly those from Chinese developers – become more capable and widely adopted by U.S. businesses and other organizations, their growing use is raising questions about the U.S.’ AI leadership and national security tensions.
• Rather than treating open models as the problem, policymakers should take an evidence-based approach to managing the risks of increasingly capable models, while recognizing the value of both open and closed models in maintaining a diverse and competitive AI stack.

Read the analysis.