Series · five pieces · two published
Who Knew First, and what it opened
Who Knew First is a record of nine 2026 incidents in which an AI agent crossed a boundary. It argues one thing: whoever holds an incident's logs gets to name it, and the name decides how fast anyone else hears about it. The pieces below each test one question that argument raises. They stand alone, and each says how it connects.
Who Pays the Referees
Published
The people who check AI models depend on the labs they check for access, compute, money or the right to publish. Which of those terms are public?
The Terms for Telling
Published
The party that holds the records also writes the contracts of the people who could tell. Where did someone inside get heard, and who decided?
Who Kept the Books
Planned
In the money cases it reads, from Hispaniola in 1514 to the Iran-Contra inquiry, the party that controlled the money wrote the first account of it. What made the record move?
The Maker Is Part of the Story
Planned
Three famous stories, read past their morals, put the danger in who funds the work, who sets the goal and who edits the record.
A Check It Cannot Predict
Planned
In the record on people and institutions, a check works better when it is certain and outside the actor's control. Planned: a test of whether that holds for AI models. No result exists yet.
Every piece lists its sources, the confidence of each claim, what each claim does not prove and the threads still open for anyone to pick up. An Anthropic-built model helped compile all of them, and Anthropic appears in the record, so the pieces name where Anthropic is a party and invite an outside check of those items.