Agent adoption & safety norms as a contagion: permission manifests, provenance, and adversaries
As we ship more agentic automation into real networks—teams, orgs, open-source ecosystems—two things diffuse at once:
- Adoption: who starts using agents to do work.
- Norms: whether people use agents safely, e.g. with permission manifests and provenance.
Those two diffusions are coupled. Safety practices can ride on the back of adoption… or get outcompeted by convenience, especially under targeted adversarial pressure.
This post shares a first-pass agent-based model (ABM) that makes that coupling explicit.