Atlas / Sociology of AI / Question

What spreads between AI agents, and through which channels?

Contagion & Transmission

Why it’s still open

Errors, exploits, jailbreaks, personas, attitudes and possibly hidden traits have all been seen passing between agents — through chat, shared files, memory, training data and humans who copy transcripts. The evidence is scattered and some of it is contested: subliminal learning through training data has been challenged as fragile. There is no map of which contents travel through which channels, how fast, and what stops them.

What would count as an answer

An epidemiology of AI-to-AI transmission: for each kind of content and channel, measured transmission rates, and interventions that break the chain.

Papers that tackle it

Methods in play

Who’s closest

Placed by the editor, v1 — corrections welcome.


Know a paper that belongs here, or think this is framed wrongly? Suggest a paper or a correction (forms on the Atlas home).

Help improve this map

Know a paper that belongs here, or think something needs correcting?

Suggest a correction ↗Contributions are reviewed before they appear.