The AI Insurance Industry

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-safety2-ai-insurance-creators

1. What is the core question insurance underwriters ask when pricing risk for any policy?

   1. What is the company's total annual revenue?
   2. Who are the company's main competitors?
   3. What profits has the company made in the last five years?
   4. What can go wrong, how often, and how badly?

2. Which of the following is an example of 'affirmative AI coverage'?

   1. A clause that limits coverage to intentional damage only
   2. A policy that excludes all AI-related losses
   3. A requirement that companies document their AI models
   4. A policy that specifically covers AI-caused errors and omissions

3. What must a company typically provide to obtain AI insurance coverage?

   1. The full source code of the AI model
   2. Model documentation, red-team reports, deployment monitoring, and human-in-the-loop controls
   3. A list of all employees who will use the AI system
   4. Proof of government approval for the AI system

4. Why is systemic risk particularly difficult to insure in AI?

   1. Insurance companies are not allowed to cover tech companies
   2. Government regulations prohibit systemic risk coverage
   3. Many firms using one foundation model could fail the same way simultaneously
   4. AI systems rarely affect multiple companies at once

5. What is 'moral hazard' in the context of AI insurance?

   1. The danger that having insurance reduces incentives to prevent harmful AI outcomes
   2. A type of fraud committed by AI developers
   3. The ethical problem of charging different premiums based on demographics
   4. The risk that AI systems will steal company funds

6. What does Munich Re's aiSure product provide?

   1. Performance insurance guaranteeing third-party AI vendor benchmarks
   2. Legal defense coverage for AI copyright lawsuits
   3. Cybersecurity protection for AI systems
   4. Life insurance for AI researchers

7. Why is 'cascading agent losses' particularly challenging for insurers?

   1. Agents rarely cause losses that affect multiple parties
   2. Insurance companies do not understand AI agents
   3. Agents are expensive to insure
   4. An autonomous AI can make many correlated bad decisions very quickly

8. What is the primary concern about 'emergent capability' as an insurable risk?

   1. Emergent behavior is always intentional
   2. It only affects small companies
   3. Insurers cannot predict or price for behaviors that appear after deployment
   4. It only occurs in academic research settings

9. What is required to prevent moral hazard in well-designed AI insurance?

   1. Companies should self-certify their safety practices
   2. Insurance should only cover intentional harms
   3. Active risk-management obligations, not just payouts
   4. Payouts should be made quickly without conditions

10. Why is copyright exposure difficult for insurers to price?

    1. Insurance companies are exempt from copyright law
    2. The legal landscape is still being litigated and creates uncertainty
    3. Copyright cases are always resolved quickly
    4. AI never generates copyrighted content

11. What typically happens to older E&O and cyber policies regarding AI losses?

    1. They are being updated to cover more AI scenarios
    2. They automatically cover all AI-related losses
    3. They are adding AI exclusions to carve out AI-caused losses
    4. They are being cancelled entirely

12. What is an 'AI performance guarantee' in insurance terms?

    1. A requirement that AI companies hire guarantee companies
    2. Insurance that covers the gap if a model underperforms a contracted benchmark
    3. Legal protection against AI performance lawsuits
    4. A promise that AI will never malfunction

13. What does the lesson identify as the 'big idea' about insurance's role in AI?

    1. Insurance only matters for physical damage, not digital
    2. AI will make insurance companies obsolete
    3. When insurers decide what is insurable, they partly decide what is deployable
    4. Insurance is becoming faster and more exciting due to AI

14. What type of risk are Lloyd's of London experiments with frontier-AI risk syndicates attempting to address?

    1. Small, individual AI errors
    2. Catastrophic risks from advanced AI systems that could cause widespread harm
    3. Routine software bugs
    4. Employee misconduct

15. Why do insurers generally exclude intentional harm and nation-state attacks from AI policies?

    1. AI cannot be used for intentional harm
    2. Intentional acts and nation-state attacks are considered uninsurable due to their nature
    3. These risks are too small to matter
    4. Such risks are predictable and easily priced