Open Source vs Closed Models
Whether AI model weights should be freely shared or kept proprietary.
An "open" AI model is one whose weights anyone can download, study, and modify. A "closed" model is only available through a company's API. The distinction matters for safety, competition, and concentration of power.
Why it matters to voters
Open models enable independent research and competition; they also lower barriers to misuse. Closed models concentrate capability in a few companies; they also make safety testing easier to enforce.
Recent developments
- 2025U.S. Commerce Department reviews export controls for the largest open-weights releases.
- 2026Industry coalitions split publicly over whether new safety rules should apply to open-weights models.
Political & policy relevance
The split runs through the AI industry itself. Leading the Future generally favors a permissive federal framework; Anthropic-aligned PACs favor more testing and disclosure regardless of openness.
Candidates working on this
Alex Bores · Sam Liccardo
Further reading
Last updated 2025
More issues