Methodology
Methodology changelog
This page records meaningful changes to formulas, sources, fallback behavior and public methodology documentation.
2026-07-02
v1.2 — two new live signals
AI incidents went live via the AI Incident Database public feed (30-day report velocity, ~60/month baseline) and a Public-attention signal was added from official Wikimedia pageviews on AI-risk articles (7-day vs 5-week baseline). Base weights rebalanced to Market 30%, AI attention 25%, Prediction markets 15%, Incidents 15%, Public attention 15%.
2026-06-27
Content-trust expansion
Added public source documentation, editorial policy pages, job-risk methodology, richer stock explanations and current-brief content surfaces.
2026-06-26
Trust and hardening pass
Added clearer demo-data notices, source/fallback language, stronger metadata and safer stock/provider behavior.
2026-06-25
Initial v1 composite
Launched the AI Fear Index as a 0-100 weighted blend of market movement, AI-news attention and prediction-market signals.