What OpenAI's industrial policy proposal reveals about who gets to draw the lines
"Standards, categories, technologies, and other infrastructural elements never speak for themselves."
Bowker & Star, Sorting Things Out
This is the fourth essay in the Invisible Infrastructure series, which studies how classification systems function as infrastructure. This essay examines a classification system being proposed in real time, by an entity with clear structural interests in where the lines get drawn.
In April 2026, OpenAI published a thirteen-page proposal for governing advanced AI. To read it well, I convened a Council of Elders: six thinkers whose frameworks are the questions this document most carefully avoids.