The Blue Whale AI Academy https://validator.w3.org/feed/docs/rss2.html The Blue Whale AI Academy – Transforming Lives Through Education Which role mainly tests AI systems by legally attacking them? You find a security weakness in a company’s chatbot. What should you do? Which standard is a certifiable management system for AI? When did the EU AI Act become broadly applicable? Why is an AI inventory important? Quiz-Rules, Ethics and Your Career Your Career in AI Security AI Rules and Governance Ethics and the Law Quiz-Watching and Responding – The SOC View Under the GDPR, a reportable personal data breach must usually be reported within… What is usually the first containment action for a hijacked agent? Which pattern is the strongest warning sign? Which log field helps you find the content that carried an injection? What does a SOC do? When Things Go Wrong: Incident Response Spotting Suspicious Behaviour Logs: The Agent’s Diary Quiz-Protecting AI Agents Defense in depth means… What is approval fatigue? Which action most clearly needs human approval? Where should an important rule like “no refunds above 50 EUR” be enforced? Which control limits damage most if an agent is completely tricked? Defense in Depth Keeping a Human in the Loop Guardrails and Checks Least Privilege Quiz-How Attackers Trick AI Agents A connector was safe at installation but became malicious after an update. This is called a… What makes memory poisoning especially dangerous? An HR bot uses an admin account to answer questions from all employees. What is the main risk? Why should passwords never be placed in a system prompt? A web page contains hidden text that tells an AI shopping assistant to recommend a scam product. What is this? Supply-Chain Attacks and MCP Too Much Power: Tool Misuse Jailbreaks and Leaked Instructions Poisoned Memory and Knowledge Prompt Injection Quiz-Security Basics for the AI Age What are the four functions of the NIST AI RMF? Which framework describes attacker tactics and techniques against AI systems? Why is untrusted input especially dangerous for language models? . An attacker floods an agent with requests so the bill becomes huge. Which goal is affected? An AI bot reveals a customer’s address to another customer. Which CIA goal failed? The Security Maps: OWASP, MITRE ATLAS, NIST Trust and Untrusted Input The CIA Triad, Explained with AI Quiz – Welcome to AI Agent Security