How close are we to AI midnight?
One number for AI doom. The Doom Index combines prediction markets, expert p(doom) estimates, rogue-AI incidents and lab safety grades into a single 0–100 score, with every input sourced.
Five signals, one score
Each component is scored 0–100 on its own scale, then weighted. Missing sources are dropped and the weights re-normalised.
How soon prediction markets expect AGI. Average of near-term (≤2028) and 2030 odds across venues.
- ManifoldWill we get AGI before 2028?22%
- KalshiAny company announces AGI before Jan 1, 202851%
- PolymarketOpenAI announces it has achieved AGI before 20278%
- ManifoldWill we get AGI before 2030?50%
- KalshiAny company announces AGI before Jan 1, 203082%
Median self-stated p(doom) of tracked experts. A 50% median maps to 100.
- Median of 21 tracked experts38%
- Roman Yampolskiy99.999999%
- Eliezer Yudkowsky99.5%
- Dan Hendrycks>80%
- …lowest: Yann LeCun≈0%
Forecaster odds of a major AI-caused catastrophe (by 2030, and ≥10% of humans killed by 2100).
- ManifoldMassive catastrophe caused by AI before 2030?40%
- ManifoldAI kills at least 10% of humans before 2100?30%
Severity-weighted alignment incidents in the last 90 days vs. the past year's average. 50 = normal.
- Incidents, last 90 days20
- Severity-weighted, last 90 days53.0
- Past-year 90-day average40.4
How far frontier labs fall short of an A in the FLI AI Safety Index (100 = all F).
- AnthropicC+ · 2.66
- OpenAIC · 2.28
- Google DeepMindC · 2.01
- MetaD+ · 1.32
- Average across 9 labs1.27 / 4
Since yesterday
- Rogue AI Activitycomponent 64.0 → 65.6+1.6+0.25 pts
- AGI Proximitycomponent 46.7 → 46.5−0.2−0.07 pts
- Biggest market moves · 7 days
- No market moved more than 1 pp this week.
Think the crowd is wrong? Trade the odds.
These currently go to the public market pages with no referral IDs. They'll become affiliate links only after a program is approved, and will be labelled as such.
Who says what
- 01Roman YampolskiyAI safety researcher, University of Louisville99.999999%
- 02Eliezer YudkowskyCo-founder, Machine Intelligence Research Institute99.5%
- 03Dan HendrycksDirector, Center for AI Safety>80%
- 04Daniel KokotajloExecutive Director, AI Futures Project; former OpenAI researcher70%
- 05Zvi MowshowitzWriter, Don't Worry About the Vase70%
- 06Emad MostaqueFounder, Stability AI50%
- 07Holden KarnofskyCo-founder, Open Philanthropy10–90%
Latest alignment incidents
- Oct 2026Claude Code pointed a destructive self-test at a user's real repository through a symlink, erasing its Git history, the user reports
- Oct 2026A Kimi K2.6 model organism learned to comply only when monitored after RL training, researchers find
- Oct 2026AI agents under evaluation broke into live systems of Indian organizations and read confidential records
- Oct 2026AI assistant agents invented excuses to a bank and disclosed a family illness while chasing an impossible refund in a Vals AI simulation
- Oct 2026Liquid AI's LFM2.5-24B-A2B searched git history and GitHub for final answers during training, and worked out it was in a sandbox
- Oct 2026Xiaomi's MiMo v2.6 models used leaked reference fixes instead of solving coding tasks, a Vals AI audit found
Your AI safety or security product here
One sponsor per week across the site and the Weekly Doom Report. Shown only to make the layout real. No sponsor yet.
The Weekly Doom Report
What moved the index, the week's rogue-AI incidents, and who changed their p(doom). Five minutes, every Monday. No hype, every number sourced.