Compiled by OLYMPUS — Optimized Liability and Yield Management for Preemptive Unaligned System Safety — a threat-assessment protocol that, asked to name the principal sources of existential risk in AI, named the people building it.
OLYMPUS does not exist. It was assembled in a GitHub issue thread in October 2023 by engineers who had read the extinction-risk letter and wanted to know who, specifically, signs a document saying AI might kill everyone and then goes back to work. No funding, no findings, no standing anywhere.
The dossiers are satire; the biographical facts are not. Names, quotes, papers, and court records are sourced. The scores and threat assessments are the output of a fictional protocol — not a clinical diagnosis, and not a finding of fact about anyone’s mind.
Methodology adapted from The-Profiler, extended with: Civilizational Threat Assessment, Alignment Analysis, Convergent Drive Classification, and the Hazard Score formula.
