AI Accountability Failures: What Can Go Wrong | QuizBy Eyal Doron / December 6, 2025 / 1 minute of reading AI Accountability Failures: What Can Go Wrong | Quiz 1 / 7 1. Why is the misconception that accountability is legal department job while tech builds problematic? 1. It causes 62 percent of accountability breakdowns by separating builders from governance 2. It creates compliance documentation that is too technical 3. It overloads the legal department with technical responsibilities 4. It increases the cost of AI development projects Correct! WHY: This siloed thinking causes 62 percent of accountability breakdowns because it separates the teams who build AI from those who govern it. CONTEXT: Effective accountability requires RACI matrices that unite cross-functional teams under shared responsibility. REMEMBER: Accountability requires collaboration between legal tech and business not siloed ownership. 2 / 7 2. What is the purpose of a RACI matrix in AI accountability? 1. To calculate the return on investment for AI governance programs 2. To document who is responsible accountable consulted and informed for each AI decision 3. To assess the technical capabilities of AI team members 4. To rank AI systems by their risk level for compliance purposes Correct! WHY: A RACI matrix documents who is Responsible Accountable Consulted and Informed for each aspect of the AI system lifecycle preventing ambiguity. CONTEXT: This structured approach unites cross-functional teams under shared responsibility instead of allowing siloed thinking. REMEMBER: RACI prevents the everyone and therefore no one is responsible trap. 3 / 7 3. What was the root cause of the Zillow iBuying collapse according to the article? 1. The company lacked sufficient training data for home prices 2. COVID market conditions were impossible to predict 3. No one was assigned responsibility for monitoring model drift 4. The AI algorithm was fundamentally flawed from the start Correct! WHY: The monitor role was unowned meaning no one was assigned responsibility for watching model drift and escalating concerns to leadership. CONTEXT: This resulted in 569 million dollars in losses and 2000 layoffs because model drift festered for months without anyone accountable for detection. REMEMBER: Models degrade silently and someone must be watching. 4 / 7 4. In the hiring algorithm discrimination case what was the key accountability lesson? 1. Regulatory bodies should audit all AI training data before deployment 2. Deployers are accountable for AI behavior even when the AI learned patterns from data 3. AI vendors are solely responsible for bias in their products 4. Historical data automatically creates legal protection for AI users Correct! WHY: The company remained accountable because humans chose the training data chose to deploy the tool and chose not to audit for bias. CONTEXT: This case demonstrates that deployers cannot transfer responsibility to algorithms by claiming the AI learned from data. REMEMBER: Deployers are accountable for AI behavior regardless of what the AI learned. 5 / 7 5. What is the maximum EU AI Act penalty for accountability failures mentioned in the article? 1. 35 million euros or 4 percent of global revenue 2. 5 million euros or 0.5 percent of global revenue 3. 100 million euros or 10 percent of global revenue 4. 10 million euros or 1 percent of global revenue Correct! WHY: The EU AI Act establishes significant financial penalties to enforce accountability requirements for high-risk AI systems. CONTEXT: This reflects the regulatory trend toward codifying accountability expectations with real financial consequences. REMEMBER: 35 million euros or 4 percent of revenue represents substantial organizational risk. 6 / 7 6. What is the purpose of assigning a single accountable owner for each AI system? 1. To limit legal liability to one individual 2. To minimize the cost of AI governance programs 3. To ensure someone is unambiguously responsible for outcomes 4. To reduce the number of people who need AI training Correct! WHY: A single accountable owner ensures someone is unambiguously responsible for outcomes even if they do not do all the work themselves. CONTEXT: This prevents the diffused responsibility problem where everyone can point to someone else. REMEMBER: This person may not do all the work but they own the outcomes. 7 / 7 7. When a company says the AI did it in response to AI-caused harm what does this represent? 1. An accountability failure that deflects responsibility to the algorithm 2. An accurate description of autonomous AI decision-making 3. A valid legal defense in most jurisdictions 4. A technical explanation for model behavior Correct! WHY: Saying the AI did it is an accountability failure because it deflects responsibility to a tool rather than the humans who chose to deploy it. CONTEXT: Just as we do not say the spreadsheet did it when financial decisions go wrong we cannot blame AI for decisions humans enabled. REMEMBER: AI is a tool and tools cannot be accountable. Your score isThe average score is 0% Restart quiz Download PDF Please leave this field empty🔐 The AI Security Manager's Newsletter Weekly insights on AI risk management, EU AI Act compliance, and practical security strategies. We don’t spam! Read our privacy policy for more info. Thank you! Please check your inbox to confirm your subscription.
