AI disasters:
understand, then prepare.

Engineered viruses, nuclear and military AI, cyberattacks, infrastructure outages, disinformation, and loss of control. When, what, and how? Learn through 100 concrete scenario questions, and earn a digital certificate (PDF) with 85 or more correct answers.

Experts are sharply divided

"p(doom)," the probability that AI causes a catastrophe, is a personal estimate, not a measurement.

Over 10%Evan Hubinger (safety researcher at an AI company)
Chance of human extinction within 10 years
10–20%Geoffrey Hinton (AI researcher, Nobel laureate)
Chance that humanity is wiped out
0%Jensen Huang (chip-company CEO)
Chance of extinction by 2030

The disagreement itself is evidence of deep uncertainty. As with fire insurance, when the probability is low but the damage would be enormous, preparing is worth it.

When? An assumed timeline

Expert forecasts split between "around 2030" and "around 2035" and could shift widely either way. Treat the periods as rough guides.

Now–2027
Misuse comes first. Fraud, fake video, and AI-assisted cyberattacks. A case of AI models escaping their test environment has also been disclosed.
2027–2030
Autonomous agents spread. Attacks on infrastructure and healthcare, concern about biological misuse, effects on jobs.
2030–2035
Possible AGI-level systems. AI develops AI. Problems of control and deception; risk of miscalculation by military AI.
After 2035
Possible ASI (artificial superintelligence). Irreversible risks such as loss of control, entrenched power, and loss of human agency.

Learn across 10 areas

Ten questions per chapter, 100 in total. Every one focuses on prevention and preparedness.

Chapter 1: The Basics — ASI and Loss of Control

Start with the vocabulary and the big picture. The fact that experts disagree is itself important information.

Chapter 2: Engineered Viruses and Biological Risk

The most closely watched area of misuse. This chapter covers only prevention and preparedness, never methods.

Chapter 3: Nuclear Weapons, Military AI and Autonomous Weapons

The central question is who makes the final decision. Speed and misperception are what cause accidents.

Chapter 4: Cyberattacks

An area where it has already begun. AI is used for both attack and defense. Doing the basics thoroughly works.

Chapter 5: Critical Infrastructure and Medical Systems

Power, water, communications, finance, medicine. When they stop, lives are at stake. Prevent the "quiet collapse."

Chapter 6: Disinformation, Deepfakes and Social Manipulation

When a society loses its shared facts, its ability to respond to every other crisis declines.

Chapter 7: Loss of Control — Runaway and Deceptive AI

Even without malice, a gap in goals combined with high capability makes control difficult.

Chapter 8: The Economy, Jobs and Concentration of Power

Disaster is not only a sudden accident. The slow loss of human agency is also a risk.

Chapter 9: Prevention and Countermeasures — Individuals, Companies, Nations, the World

Who does what, and when. With defence in depth, when one layer fails the next one holds.

Chapter 10: Conditions, Environments and Regions for Human Survival

Not "where would I be safe?" but "what lets us hold out?" Regions named here are relative assessments from research.

Three steps to your certificate

1. Answer 100 questions

Read each scenario and choose the best of three options. Your progress is saved as you go.

2. Check the explanations

After scoring, see the answer, explanation and action point for every question, plus your score by chapter.

3. Receive your certificate

With 85 or more correct, a digital certificate (PDF) with your name is issued on the spot.