EU AI Act Quiz

Answers and explanations appear as you take the quiz.

1. What organizing principle does the EU AI Act use to set its rules?
   • A risk-based tier system based on how the AI is used
   • The size of the company building the AI
   • Whether the AI uses neural networks or not
   • The country where the AI model was trained
2. Which of these is an 'unacceptable risk' practice banned outright by the Act?
   • Using AI to recommend movies
   • Government social scoring of citizens
   • A chatbot answering customer questions
   • AI that filters spam email
3. For an AI system covered by Article 50 that interacts directly with people, such as a chatbot, what disclosure applies?
   • Pass a full conformity assessment
   • Register in an EU-wide database
   • Disclose to users that they are interacting with AI
   • Submit to adversarial red-team testing
4. What maximum administrative fine does the Act set for violating the prohibition on specified AI practices?
   • Up to €15,000 or 1% of turnover
   • There are no financial penalties
   • Exactly €1 million in every case
   • Up to €35 million or 7% of worldwide annual turnover for an undertaking, whichever is higher
5. How does the Act treat the same AI model used in two different products?
   • Classification can differ because risk depends on the use case
   • It always gets the same classification everywhere
   • Only the first product is regulated
   • Models are exempt regardless of use
6. Which extra duty applies to the most capable general-purpose 'systemic risk' models?
   • They must be open-sourced
   • Model evaluations and adversarial testing are required
   • They are banned in the EU
   • They face no special requirements
7. Which set of obligations applies to providers of general-purpose AI models under the EU AI Act?
   • Mandatory biometric identification of all users
   • A ban on publishing any model documentation
   • Technical documentation, a copyright policy and a public summary of training content
   • Automatic approval for every downstream system
8. Which extra obligations does the Commission list for GPAI models with systemic risk?
   • Only disclose that a chatbot is AI
   • Only register a company domain
   • Waive all documentation and testing
   • Notify the Commission, assess and mitigate risks, report incidents and provide cybersecurity protections