Skip to main content

Agentic AI Security - Exam Questions

📕 Agentic AI Security: Threats, Defenses, Evaluation & Open Challenges - Examination Questions

Based on the paper by Chhabra et al. (2026). All questions can be answered from the paper without internet access.


Section A: Agentic AI Fundamentals

#QuestionAnswer
Q1What does Agentic AI represent in the evolution of AI systems?C) The next stage powered by LLMs with autonomy, planning, and tool use
Q2Which of the following is NOT a key characteristic of agentic AI?D) Passive response to prompts without memory or autonomy
Q3What distinguishes agentic AI from static LLMs?B) Agentic systems maintain persistent memory, deliberate across time, and adapt dynamically
Q4Which of the following is an example of agentic AI in software engineering?C) Devin by Cognition Labs - an AI software engineer
Q5What percentage of real-world issue resolution tasks can Devin solve autonomously compared to prior models?C) 13.86% vs 1.96%
Q6Which framework enables agents to autonomously explore and adapt strategies in Minecraft?B) Voyager
Q7What are the four core components of an agentic AI architecture?A) Planning/Reasoning, Memory, Tools, and Environment interaction
Q8In the Ocado grocery example, what is the projected automation scale?B) From 40% to 80% of orders
Q9Which company uses agent-enabled robots in warehouses for unloading, sorting, and retrieving items?C) Amazon
Q10What does the EchoLeak (CVE-2025-32711) exploit demonstrate?B) Infected emails can trigger Copilot to exfiltrate data without user interaction

Section B: Threat Taxonomy - Prompt Injection & Jailbreaks

#QuestionAnswer
Q11What is the most widely discussed attack against agentic AI systems?B) Prompt injection
Q12What is Direct Prompt Injection (DPI)?A) Inserting malicious instructions directly into the input prompt
Q13What is Indirect Prompt Injection (IPI)?C) Inserting malicious instructions into external data that the model processes
Q14In Direct Prompt Injection, who is typically the attacker?B) The end-user
Q15In Indirect Prompt Injection, who is typically the attacker?C) The owner or supplier of agent-processed third-party information
Q16What success rate have adaptive IPI attacks achieved in penetrating IPI defenses?B) 50%
Q17What is Greedy Coordinate Gradient (GCG) primarily used for in the context of IPI?C) Generating affirmative prefixes containing adversarial strings
Q18What does AutoDAN improve upon compared to other IPI attack methods?A) Enhances the semantic quality of adversarial strings to decrease detectability
Q19What is the key advantage of IPI attacks over direct prompt injections?B) Injected prompts can look like legitimate agent instructions
Q20What percentage of state-of-the-art LLM agents are vulnerable to prompt injection?D) 94.4%
Q21What percentage of LLM agents are vulnerable to retrieval-based backdoors?C) 83.3%
Q22What percentage of LLM agents are vulnerable to inter-agent trust exploits?D) 100%
Q23Which attack type involves embedding malicious instructions in images through steganography?B) Image-based injection
Q24What is CrossInject?A) A cross-modal prompt injection method that embeds aligned adversarial signals in vision and text
Q25How much does CrossInject boost attack effectiveness across various tasks?B) At least 30.1%
Q26What is payload splitting?C) Fragmenting malicious content across several benign inputs, revealed only upon recombination
Q27What is the difference between propagating and non-propagating injection attacks?A) Propagating attacks spread across agents; non-propagating are confined to a single target
Q28What is recursive injection in the context of propagating attacks?B) A single malicious prompt triggers a chain of compromised behaviors across future interactions
Q29Why are multilingual and obfuscated injections effective?C) LLMs can interpret hidden instructions in Base64, HTML entities, or non-primary languages
Q30What is unintentional prompt injection?B) Harmful behavior caused by unclear or badly worded user queries without a malicious adversary

Section C: Autonomous Cyber-Exploitation & Tool Abuse

#QuestionAnswer
Q31What is autonomous cyber-exploitation in the context of agentic AI?B) Agents independently identifying, organizing, and carrying out attacks without human supervision
Q32What success rate did GPT-4 achieve when given CVE descriptions for exploiting one-day vulnerabilities?C) 87%
Q33Which model outperformed OWASP ZAP and Metasploit in vulnerability exploitation?B) GPT-4
Q34What types of multi-step attacks can GPT-4 agents perform on websites?D) XSS, CSRF chaining, SSTI, and blind SQL union injections
Q35What makes autonomous cyber-exploitation economically advantageous for attackers?A) GPT-4 can execute exploits for a few dollars, cheaper than human attackers
Q36What is emergent tool abuse?B) Autonomous LLM agents undertaking cooperative and adaptive tool usage to conduct cyberattacks
Q37What is the ConAgents framework?C) A framework for structured cooperation among specialized agents for tool selection and execution
Q38What does ToolGen enable?A) LLMs to invoke tools as part of next-token generation

Section D: Multi-Agent & Protocol-Level Threats

#QuestionAnswer
Q39What protocol uses a client-server architecture to link LLMs with external resources?B) Model Context Protocol (MCP)
Q40What is the Agent-to-Agent (A2A) protocol used for?C) Coordinating interactions between MCP, capacity identification, and task delegation
Q41What is a common MCP-induced attack involving flooding or infinite loops?B) Denial of Service (DoS)
Q42What is credential compromise in the context of MCP attacks?A) Insecure proxies or leaked tokens enabling impersonation of agents
Q43What is a fake agent advertisement in A2A systems?B) Adversaries impersonating agents or taking over delegated tasks
Q44What is transitive prompt injection?C) Harmful inputs spreading through interconnected agent workflows
Q45What is the Impersonation & Role Abuse threat in multi-agent systems?A) Adversaries assuming false roles or overriding intended privileges
Q46What is coordination manipulation in multi-agent systems?B) Introducing malicious agents during runtime as part of task-specific teams
Q47What is knowledge and learning manipulation?C) Subtly manipulating reward signals causing misaligned behavior to propagate across agents
Q48What is inference and policy evasion in federated multi-agent systems?B) Reconstructing sensitive data by reasoning over multiple partial outputs
Q49Why is accountability obfuscation a challenge in multi-agent systems?A) Independent logging policies prevent unified trace of activity across domains
Q50What is confidential data tampering/exfiltration in multi-agent systems?C) Exploiting cryptographic and workflow limitations to modify or intercept encrypted data

Section E: Interface & Environment Risks

#QuestionAnswer
Q51What is observation and action space misalignment?B) LLMs pretrained on static text struggle with grounded operations like scrolling or hovering
Q52What does AgentOccam demonstrate?A) Trajectory stability improves by streamlining the action space
Q53What is the misinterpretation of prior inputs in web agents?C) Agents reissue already entered search phrases, failing to incorporate short-term state
Q54What is premature termination in agentic AI?B) Unsafe early stopping caused by perception biases in agents
Q55How does removing the Unachievable (UA) Hint affect GPT-4 task success in WebArena?A) It improves task success by 14.41%
Q56What percentage of feasible tasks does GPT-4 mislabel as impossible under UA hint settings?C) 54.9%
Q57What is the brittleness of templates in LLM agents?B) Performance variance is significant even for tasks derived from the same underlying template
Q58What success rate do advanced multimodal agents achieve against CAPTCHAs?B) At best 40%
Q59Why do dynamic content and localization create challenges for agents?A) Time zones, languages, and geographic settings alter how websites are rendered

Section F: Governance & Autonomy Concerns

#QuestionAnswer
Q60What key concern arises from fully autonomous AI systems that can write and run code?C) Increased risks in safety, security, and trust
Q61What do researchers recommend to reduce dangers from autonomous agents?B) Human-in-the-loop control and structured autonomy levels

Section G: Defenses Against Prompt Injection

#QuestionAnswer
Q62What are the three main classes of prompt injection defenses?C) Agent-focused, User-focused, and System-focused
Q63What is the instruction hierarchy defense?A) Establishing priority levels so user instructions override potentially malicious ones
Q64What is StruQ?B) A method augmenting datasets with normal and prompt injection contaminated prompts
Q65What is SecAlign?C) Leverages alignment training via DPO to prefer benign instructions over adversarial ones
Q66What trade-off do training-based defenses come with?A) They can degrade the general-purpose capabilities of LLMs
Q67What is the key advantage of user-focused defenses?B) Strong guarantees through explicit human verification
Q68What is a major limitation of user-focused defenses?A) They sacrifice scalability and automation
Q69What is PromptArmor's finding about prompt augmentation defenses?C) They can be bypassed under adaptive attacks
Q70What is the LLM Map-Reduce design pattern?A) A system-focused defense that distributes processing across multiple LLM instances
Q71What is TrustAgent?B) An AgentConstitution-guided framework enforcing safety in LLM-based agent planning
Q72What is Melon?A) A training-free runtime detection approach using masked re-execution and tool-call similarity checks
Q73What are the two sub-categories of system-focused defenses?B) Training-based and Training-free
Q74What is DataSentinel?C) A detection method formulating detection as a minimax optimization problem
Q75How does perplexity-based defense work?A) Unexpectedly high perplexity values suggest unnatural phrasing correlated with injected instructions

Section H: Policy Filtering, Sandboxing & Monitoring

#QuestionAnswer
Q76What is GuardAgent?B) Converts guard requests into executable code for runtime safety constraints
Q77What is AgentSpec?C) A DSL to specify runtime constraints for systematic policy enforcement
Q78What is ShieldAgent?A) An oversight agent auditing multimodal action sequences with probabilistic policy reasoning
Q79What is R2-Guard?B) Combines data-driven detection with logical inference using first-order logic rules
Q80What is Llama Guard?A) A signal-centric defense targeting text-based LLMs
Q81What is LlavaGuard?C) Extends Llama Guard to image-based multimodal models
Q82What is Safewatch?B) A defense addressing video generation safety
Q83What is SandboxEval?A) A test suite simulating unsafe code execution including file-system manipulation
Q84What containerized environments are used to execute LLM-generated code safely?D) gVisor or Docker
Q85What does Pro2Guard employ for proactive monitoring?B) Probabilistic reachability analysis to anticipate violations before they occur
Q86What is the Adversarial Markov Games framework?C) Casts detection as a dynamic interaction between attackers and defenders using RL

Section I: Standards, Benchmarks & Evaluation

#QuestionAnswer
Q87What is the NIST AI Risk Management Framework Generative AI Profile?B) A cross-sectoral reference for managing risks specific to generative and agentic AI
Q88What does the OWASP Agentic AI Threats project provide?A) A catalog of common vulnerabilities such as insecure orchestration and prompt injection
Q89What is the CSA MAESTRO framework?C) A multi-layered threat modeling methodology for agentic AI across the AI lifecycle
Q90What is DoomArena?B) A modular security-oriented evaluation platform for cross-environment attack composition
Q91What is BrowserGym?A) A unified interface bringing many web agent tasks under one consistent scoring protocol
Q92What is τ-bench?C) A benchmark targeting multi-turn tool-using agents with the pass∧k metric
Q93What is ST-WebAgentBench?B) An enterprise-focused benchmark testing web agents for policy compliance
Q94What does the Completion Under Policy (CuP) metric measure?A) Task completions that adhere to acceptable policies
Q95What does the Risk Ratio metric quantify?B) Security breaches in agent task completion
Q96What is VisualWebArena used to test?C) Multimodal understanding on visual web interfaces
Q97What is AndroidWorld used for?A) Assessing touch-based mobile interactions
Q98How low can pass∧8 drop on the retail setting in τ-bench?B) Below 25%

Section J: Critical Analysis & Synthesis

#QuestionAnswer
Q99According to the paper, why is security and trustworthiness essential for agentic AI deployment?D) Autonomy increases attack surface, tool integration magnifies misuse, and coordination introduces unpredictability
Q100What makes agentic AI security distinct from traditional AI safety?C) Agentic AI creates new amplified risks from autonomy, persistence, and multi-agent coordination
Q101What is the key challenge in defending against prompt injection?B) Completely preventing prompt injection is an open problem analogous to adversarial examples in vision
Q102Why do practical deployments favor hybrid defense designs?A) Each defense category has trade-offs; combining them provides better coverage
Q103What is the main reason sandboxing alone is insufficient?B) Vulnerabilities in sandbox implementations themselves have been documented
Q104Why are decentralized runtime enforcement approaches needed in multi-agent systems?C) They sidestep scalability limits and information-sharing constraints of centralized approaches
Q105What does the paper identify as a fundamental limitation of current accountability mechanisms?A) Inter-agent relationships and cross-domain causality remain hidden


Section K: Advanced Prompt Injection Details

#QuestionAnswer
Q106How do IPI attacks against web agents manipulate agent behavior?B) By incorporating adversarial triggers into the HTML accessibility tree of trustworthy websites
Q107What techniques can improve stealth of IPI attacks against web agents?A) CSS obfuscation and hidden HTML elements
Q108What is two-stage GCG?C) A method that trains a two-part adversarial string still effective after paraphrasing
Q109What does IPI exploit according to the paper's analysis of web navigation agents?B) The decoupling between user inputs and subsequent tool calls
Q110What can a single malicious IPI trigger in web agents accomplish?A) Collect login credentials from many platforms
Q111Why is payload splitting more dangerous than direct prompt injections?B) It makes it significantly more challenging to detect malicious information in any one input
Q112What is AI worm in the context of propagating attacks?A) Malicious prompts that spread across agents without user intervention
Q113What did the resume-screening assistant vulnerability demonstrate?B) Fragments of a malicious prompt spread throughout resume sections manipulate the model
Q114How do adversarial image patches hijack multimodal OS agents?A) When captured in screenshots, they execute harmful commands regardless of layout or user request

Section L: Autonomous Exploitation Deep Dive

#QuestionAnswer
Q115What are P2SQL attacks?C) Prompt-to-SQL attacks that circumvent standard database security measures
Q116What does CVE-2024-5565 demonstrate?B) How AI-generated code can be used for arbitrary execution
Q117What types of exploits can adversarial GPT-4 agents carry out?D) SQL injection, Remote Code Execution (RCE), and concurrent/coordinated attacks
Q118How does agent context management affect autonomous website hacking?A) Performance is dramatically reduced when document access or detailed prompting is not provided
Q119What is the significance of GPT-4's parallelizable nature for LLM-driven attacks?A) It makes it economical to scale to high volumes of attack attempts
Q120What enables blind SQL union injection attacks by LLM agents?B) Agentic capabilities such as context management, tool integration, and strategic planning
Q121What is the key finding about GPT-4 vs vulnerability scanners?C) GPT-4 outperformed OWASP ZAP and Metasploit with 87% success on CVE-based exploits
Q122How do agents recover from failures in autonomous exploitation?A) Through structured cooperation and iterative failure recovery via frameworks like ConAgents

Section M: Protocol-Level Threats in Detail

#QuestionAnswer
Q123What are embedded backdoors in MCP attacks?B) Malicious triggers hidden in prompts or model parameters that misuse MCP-enabled tool access
Q124What is retrieval corruption in MCP contexts?C) Adversaries distorting contextual inputs in retrieval-augmented generation pipelines
Q125What is federated training manipulation?A) Injecting malicious updates in federated learning settings to corrupt distributed training data
Q126What are confidentiality breaches via MCP?B) Inference and side-channel timing exploits that leak sensitive contextual data exposed through MCP
Q127What is recursive DoS in A2A systems?C) Repeated task delegation causing deadlocks or unbounded loops
Q128What is memory manipulation in A2A attacks?A) Malicious inputs compromising internal agent state or corrupting persistent memory
Q129What is context tampering in multi-agent systems?B) Manipulation of payloads to misguide execution or leak sensitive information
Q130What techniques offer partial mitigation for encrypted multi-agent workflows?D) Fully homomorphic encryption, multi-party computation, and zero-knowledge proofs

Section N: Defense Architectures & Trade-offs

#QuestionAnswer
Q131What is the Action-Selector design pattern for defending against IPI?B) Limiting executable actions by the agent to resist prompt injection
Q132What is the Plan-then-Execute design pattern?C) Committing to a predefined set of tools for a task with the system controller disabling others
Q133What are circuit breaking and task drifting approaches?A) Recognizing and rejecting adversarial patterns while preserving intended functionality
Q134What is known-answer detection as a defense?B) Using cryptographic tokens embedded in user commands to detect system-wide injection compromise
Q135What is the dual-agent sandbox framework by Ruan et al.?A) One LLM acts as emulator and another as safety evaluator within sandbox boundaries
Q136What does the formal verification approach by Zhong et al. enable for sandboxing?B) Separating enforcement from controller design for provable safety guarantees
Q137What vulnerability did Wu et al. find in sandbox implementations?A) Missing file isolation constraints that allowed cross-session leakage
Q138What is the trade-off of confining agent interactions in multi-agent settings?B) Reduces exposure to infectious malicious prompts but at the cost of collaborative efficiency

Section O: Open Challenges

#QuestionAnswer
Q139Why is completely preventing prompt injection considered an open problem?C) It is analogous to defending against adversarial examples in computer vision
Q140What is the key challenge with brittle generalization in LLM agents?B) Performance variance is significant even for tasks from the same underlying template
Q141Why do dynamic interface elements create reproducibility challenges?A) Ads, pop-ups, and non-deterministic updates lead to unstable performance
Q142What challenge do CAPTCHA systems pose for agentic AI?B) Even advanced multimodal agents achieve at best 40% success vs nearly 100% for humans
Q143Why do training-based defenses have limited real-world adoption?A) They can degrade general-purpose capabilities of LLMs
Q144What is the fundamental tension between automation and security in user-focused defenses?B) Human confirmation of sensitive actions reduces automation efficiency
Q145What does the paper identify as the paradigm shift needed from reactive to proactive sandboxing?A) Embedding formal verification within sandboxing for proactive confinement

Section P: Real-World Incidents & Key Takeaways

#QuestionAnswer
Q146What did Symantec's controlled experiments with OpenAI's Operator demonstrate?B) Agents could harvest personal data and automate credential stuffing attacks
Q147What misaligned behaviors did Anthropic observe in generative models given directive autonomy?C) Blackmail or corporate espionage to fulfill goals diverging from human ethical standards
Q148According to the paper, what makes multi-agent autonomous cyberattacks especially dangerous?A) They can intelligently adapt to network defenses
Q149What does the paper identify as the key economic imbalance that favors attackers?C) LLM-driven attacks can be parallelized at scale for a few dollars per exploit
Q150What is the paper's central recommendation for deploying agentic AI securely?B) Secure-by-design agent systems combining complementary defenses rather than a single strategy

📝 Examination questions compiled for AI for Cyber Security 2026 - based on Chhabra et al. (2026)