Who Is Liable When AI Causes Harm? Quiz

Answers and explanations appear as you take the quiz.

1. What did the tribunal decide in Moffatt v. Air Canada?
   • The chatbot was a separate legal entity responsible for its own statements
   • The customer should have ignored the chatbot
   • Air Canada was responsible for the chatbot's incorrect information
   • The chatbot's developer alone was liable
2. What does strict product liability mean?
   • The injured person does not have to prove the manufacturer was at fault, only the defect, damage and causal link
   • The manufacturer is never liable
   • Only criminal penalties apply
   • The user must prove intent to harm
3. How does the revised EU Product Liability Directive treat software?
   • It excludes all software
   • It covers only embedded firmware
   • It covers only open-source code
   • It explicitly treats software, including AI systems, as a product
4. What happened to the proposed EU AI Liability Directive?
   • It entered into force in 2024
   • It was dropped after the Commission said in February 2025 it expected no agreement
   • It merged into the GDPR
   • It became a US federal law
5. Which new type of damage does the revised EU directive cover?
   • Destruction or corruption of non-professional data
   • Lost stock market profits
   • Reputational harm to competitors
   • Fines from regulators
6. In Mata v. Avianca, who was held responsible for fake citations produced by ChatGPT?
   • OpenAI
   • The court clerk
   • The opposing party
   • The lawyers who filed the brief
7. Under the revised EU directive, what happens when someone substantially modifies a product?
   • All liability disappears
   • Only the original manufacturer remains liable
   • The modifier can be treated as a manufacturer
   • The product is automatically banned
8. How does the revised EU directive help claimants with complex AI systems?
   • It requires AI to testify
   • It lets courts order evidence disclosure and presume defect or causation when proof is excessively difficult
   • It sets a fixed payout for every AI harm
   • It removes the need to show any damage