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. What is the purpose of a RACI matrix in AI accountability? 1. To calculate the return on investment for AI governance programs 2. To assess the technical capabilities of AI team members 3. To document who is responsible accountable consulted and informed for each AI decision 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. 2 / 7 2. What was the root cause of the Zillow iBuying collapse according to the article? 1. COVID market conditions were impossible to predict 2. The company lacked sufficient training data for home prices 3. The AI algorithm was fundamentally flawed from the start 4. No one was assigned responsibility for monitoring model drift 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. 3 / 7 3. In the hiring algorithm discrimination case what was the key accountability lesson? 1. Deployers are accountable for AI behavior even when the AI learned patterns from data 2. Historical data automatically creates legal protection for AI users 3. Regulatory bodies should audit all AI training data before deployment 4. AI vendors are solely responsible for bias in their products 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. 4 / 7 4. 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. 10 million euros or 1 percent of global revenue 4. 100 million euros or 10 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. 5 / 7 5. Why are documentation gaps considered critical accountability failures? 1. They create security vulnerabilities in production systems 2. They increase storage costs for AI systems 3. They make it impossible to reconstruct what happened and demonstrate accountability 4. They slow down model training processes Correct! WHY: Without records of decisions made throughout the AI lifecycle it becomes impossible to reconstruct what happened or explain why. CONTEXT: When regulators or litigants ask why did the AI do that the answer we do not know is itself an accountability failure. REMEMBER: Documentation is the infrastructure of accountability. 6 / 7 6. What characterizes diffused responsibility as an accountability failure pattern? 1. Responsibility split across teams so each can blame others when problems occur 2. A single executive taking too much control over AI decisions 3. Documentation spread across multiple systems making it hard to find 4. Technical complexity that prevents anyone from understanding the system Correct! WHY: When responsibility splits across multiple teams each team can point to decisions made by others creating gaps where harm falls through organizational cracks. CONTEXT: This is the everyone is responsible therefore no one is responsible trap that paralyzes incident response. REMEMBER: If everyone is responsible nobody is responsible. 7 / 7 7. What is an AI accountability failure? 1. When machine learning training takes too long 2. When responsibility for AI decisions is diffused or deflected so no one is clearly responsible 3. When AI models exceed their computational budget 4. When AI systems fail to produce accurate predictions Correct! WHY: An AI accountability failure occurs when no one can be identified as responsible for AI decisions and their consequences. CONTEXT: This creates accountability gaps where harm goes unaddressed and remediation becomes impossible. REMEMBER: If nobody is accountable then nothing gets fixed. Your score isThe average score is 0% Restart quiz Download PDF Please leave this field empty๐ The AI Security Manager's Newsletter Weekly insights on AI risk management, EU AI Act compliance, and practical security strategies. We donโt spam! Read our privacy policy for more info. Thank you! Please check your inbox to confirm your subscription.
