Free AAISM Practice Test Questions and Answers (2026) | Cert Empire

Free preview: 20 questions.

Isaca AAISM

Q: 1

As organizations increasingly rely on vendors to develop AI systems, which of the following is the MOST effective way to monitor vendors and ensure compliance with ethical and security standards?

Options

Discussion

Reese G. Jul 20, 2026 7:43 pm

Option A makes sense. Regular audits actually verify that vendors follow ethical and security standards, which is stronger than just letting them self-report or share docs. Pretty sure this is what most orgs would do in practice, but open to other thoughts.

Karan Q. Jul 15, 2026 5:39 pm

Option A is right since audits actually verify what the vendor is doing, not just taking their word for it. Self-attestation and just sharing docs aren’t enough for real compliance. Pretty sure this is what ISACA expects here.

Nathan P. Jul 10, 2026 5:56 pm

Option A is the strongest here. Regular audits actually verify vendor compliance, while options like D (self-attest) put too much trust in the vendor. Think audits are pretty standard for critical AI systems. Open if anyone has a different take.

Reese P. Jul 25, 2026 2:38 pm

Makes sense to pick A since audits directly check what vendors are actually doing, not just what they say. Self-attestation like in D feels too weak for real assurance. I think A's the best bet but open if anyone sees it differently.

RileyL Jul 19, 2026 7:38 pm

Probably A, because only regular audits give you independent evidence that vendors are following your AI security and ethics policies. Self-monitoring or self-attestation (like in B or D) relies too much on trust, so not as strong. Audits catch issues you might otherwise miss. Pretty sure that's the ISACA logic here but I'm open to debate!

Sean G. Jul 12, 2026 2:38 pm

C or A, saw similar question in a practice set and both seemed possible but A is probably expected for "most effective".

Hannah O. Jul 16, 2026 2:21 pm

For me, A. Official ISACA guides and some practice exams mention audits as the best compliance check.

Ishaan O. Jul 12, 2026 6:04 am

A for sure. Regular audits (A) are actually checking the vendor's controls, not just taking their word like D. B is too hands-off and basically trust-based, which doesn't fly for real compliance. Could be wrong but most audits catch what self-attestation misses.

Ethan I. Jul 13, 2026 8:07 pm

Probably A, seen this asked in a similar way on practice. Audits give real assurance where self-reporting just isn't enough for compliance.

Reese R. Jul 20, 2026 1:14 am

It’s A, not D. Self-attest (D) is way less reliable for actual compliance monitoring.

Be respectful. No spam.

Correct Answer:

Explanation

Conducting regular, independent audits is the most effective method for monitoring vendor compliance. Audits provide direct, verifiable assurance that the vendor's processes, controls, and outputs align with contractually mandated ethical and security standards. This approach moves beyond trust-based models like self-attestation, offering a structured and evidence-based mechanism to assess the entire AI development lifecycle. It allows the contracting organization to proactively identify gaps, enforce accountability, and ensure that principles of fairness, transparency, and security are being actively implemented, rather than just claimed.

Why Incorrect

B. Requiring vendors to monitor their adherence to ethics and security standards: This approach lacks independent verification and creates a potential conflict of interest, making it less reliable than an external audit.

C. Mandating that vendors share source code and AI documentation with the contracting party: While useful for due diligence, code and documentation review is a static, point-in-time assessment and does not guarantee ongoing process compliance.

D. Allowing vendors to self-attest ethical AI compliance and implement benchmark monitoring: Self-attestation is the weakest form of assurance, as it is not independently verified and may not accurately reflect actual practices.

References

1. ISACA. (2023). Artificial Intelligence Audit Toolkit. In the section on "Third-Party Management

" the toolkit emphasizes the need for organizations to "Perform audits of the third party" and "Review third-party attestation reports (e.g.

SOC 2)" to gain assurance over the controls at the vendor. This directly supports the practice of conducting audits (Option A) over relying on vendor self-monitoring or attestation (Options B and D).

2. ISACA. (2021). Auditing Artificial Intelligence. This white paper states

"The audit and assurance professional should obtain an understanding of the third-party relationships and related processes to determine whether the enterprise has implemented appropriate governance and monitoring controls." It highlights the auditor's role in verifying these controls

which is the essence of an audit. (Page 18

Third-Party Management).

3. National Institute of Standards and Technology (NIST). (2023). AI Risk Management Framework (AI RMF 1.0). In the "Govern" function

section 4.4 "AI System and Third-Party Relationships

" the framework outlines the importance of establishing processes for "monitoring and reviews of third-party providers

" including "audits

and assessments." This underscores that active

independent verification like audits is a core component of managing third-party AI risk. (Page 21).

Q: 2

During the creation of a new large language model (LLM), an organization procured training data from multiple sources. Which of the following is MOST likely to address the CISO's security and privacy concerns?

Options

Discussion

Adam A. Jul 8, 2026 4:54 pm

B. Data minimization.

Luke Jul 5, 2026 10:13 pm

I’d say it's B. Minimizing the data collected is the main protection for both privacy and security concerns in this scenario.

Noah Jul 15, 2026 5:11 am

Maybe C, since you need to label the data before you can focus on privacy risks. Not 100% sure.

Ravi U. Jul 25, 2026 11:47 am

Honestly, these questions always overcomplicate things. Minimization is what CISOs actually care about for privacy, not just labeling stuff. B.

Grace S. Jul 18, 2026 5:02 pm

B , since minimizing the data set actually removes sensitive info that could leak out of the LLM later. Classification helps control it but doesn't prevent risky data from getting in. Seen similar wording on practice questions, and B fits best unless they're asking for just an inventory, which they're not. Open to other views but this is pretty clear to me.

Vikram Jul 21, 2026 11:31 pm

Feels like it's B. Only minimizing the data actually deals with privacy risks up front, not just labeling or discovering it.

CalmSec867 Jul 18, 2026 6:25 pm

I get why B is tempting but C makes more sense if you want to actually address privacy up front. C.

Hannah Jul 10, 2026 8:00 pm

Ryan D. Jul 24, 2026 8:22 am

Its B, data minimization. That actually reduces the amount of sensitive data in training sets, directly cutting down privacy risk. C is a common distractor but just labeling doesn't fix exposure. Pretty sure B is what CISOs want here.

Karan G. Jul 22, 2026 5:09 pm

C tbh, since classification has to come first when pulling from multiple sources. Can't minimize what you haven't labeled yet. Pretty sure that's how most orgs would approach security here, but open to other thoughts.

Be respectful. No spam.

Correct Answer:

Explanation

Data minimization is a core privacy and security principle that involves processing only the data that is absolutely necessary for a specific purpose. In the context of training a large language model (LLM), applying data minimization means proactively identifying and removing any sensitive, personal, or proprietary information from the training datasets before the model is trained. This directly addresses the CISO's concerns by fundamentally reducing the risk surface. By ensuring sensitive data is not included, the organization mitigates the possibility of the model memorizing and later exposing this information, thus preventing potential data breaches and ensuring compliance with privacy regulations.

Why Incorrect

A. Data augmentation: This technique is used to increase the volume of training data to improve model accuracy and generalization, not to address security or privacy risks.

C. Data classification: This is a foundational step to identify sensitive data, but it does not, by itself, mitigate the risk; it only categorizes it.

D. Data discovery: This process locates data across systems. Like classification, it is a preliminary step that identifies the problem but does not implement a solution.

References

1. National Institute of Standards and Technology (NIST). (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). (NIST AI 100-1).

Page 29

Table 7

MAP.T10: In the section on documenting training data

this subcategory explicitly lists "adherence to data minimization principles" as a key element for managing AI risks. This directly supports using data minimization to handle training data concerns.

2. ISACA. (2023). Artificial Intelligence: An Audit and Assurance Framework.

Page 21

Section 3.2.1 Data Governance: The framework states

"Data used to train AI models should be relevant

accurate and appropriate for the intended purpose... It is important to ensure that the data used for training does not contain any sensitive or confidential information that could be inadvertently exposed by the AI model." This aligns with the principle of data minimization to remove unnecessary sensitive data.

3. European Union. (2016). General Data Protection Regulation (GDPR).

Article 5(1)(c): This article establishes "data minimisation" as a core principle of data protection

stating that personal data shall be "adequate

relevant and limited to what is necessary in relation to the purposes for which they are processed." This legal and privacy principle is the direct solution to the CISO's concerns when applied to AI training data.

Q: 3

An organization needs large data sets to perform application testing. Which of the following would BEST fulfill this need?

Options

Discussion

Maya H. Jul 19, 2026 7:45 am

C . Open-source data repositories literally exist to provide large, ready-to-use datasets. D is more for creating variation or synthetic data, not meeting the base requirement for large sets. Seen similar logic referenced in a few practice questions.

Ben N. Jul 8, 2026 10:23 am

Ishaan O. Jul 18, 2026 6:35 pm

Probably C, unless the question is about generating unique test data instead of sourcing existing sets?

Jordan H. Jul 21, 2026 5:08 pm

Probably C for this one.

Mason Jul 17, 2026 10:41 pm

Nah, I think C makes more sense than D. Data augmentation (D) is useful if you already have a dataset and need to expand it, but for just getting large data sets fast, open-source repositories (C) are built for this. Pretty sure that's what the exam wants, but open to debate.

QuickMentor2595 Jul 10, 2026 5:00 am

C/D? I was thinking D since data augmentation can create even more data from what you have, which helps with scale in testing. But not sure if that matches the 'best fulfill' angle since you're starting from an existing set. Maybe someone who worked with both methods can weigh in.

Alex B. Jul 13, 2026 4:37 pm

Not B, C. Saw a similar question in an exam report, open-source repo fits best for large data.

Amelia Jul 19, 2026 8:36 am

Pretty clear it's C here.

Ben S. Jul 8, 2026 6:22 pm

Makes sense, I think C is the one here.

Ravi Jul 24, 2026 10:50 pm

C tbh, D looks tempting for expansion but the question just wants large datasets, not synthetic ones.

Be respectful. No spam.

Correct Answer:

Explanation

Open-source data repositories are centralized platforms (e.g., Kaggle, UCI Machine Learning Repository, Google Dataset Search) that host a wide variety of large, often cleaned and labeled, datasets. They are specifically designed to provide data for research, development, and testing of applications, particularly in the AI and machine learning domains. Using these repositories is the most direct, efficient, and common method for an organization to acquire the large-scale data needed for comprehensive application testing without the significant overhead of collecting and preparing the data from scratch.

Why Incorrect

A. AI model cards are documents that provide transparency about a model's performance, limitations, and ethical considerations, not a source of raw data.

B. Incorporating data from search content is impractical; it is often unstructured, may have copyright or privacy restrictions, and requires extensive effort to scrape, clean, and label.

D. AI data augmentation is a technique to artificially increase the size of an existing dataset. It cannot be used without an initial dataset to augment.

References

1. ISACA

Artificial Intelligence Audit and Assurance Framework

2023. In the "AI Data Life Cycle" section

the framework discusses "Data Sourcing and Acquisition

" emphasizing the need to obtain relevant and sufficient data for training and testing AI models. Open-source repositories are a primary means of fulfilling this requirement for large-scale data needs.

2. Ng

A. (2023). Machine Learning (CS229) Course Materials. Stanford University. Lecture notes and project guidelines frequently reference the use of public datasets from repositories like the UCI Machine Learning Repository and Kaggle as standard practice for obtaining data for model development and evaluation. (See

for example

project guidelines which list acceptable data sources).

3. Roh

Heo

& Whang

S. E. (2021). A survey on data collection for machine learning: a big data - AI integration perspective. IEEE Transactions on Knowledge and Data Engineering

33(4)

1328-1347. This paper discusses various data collection methods

highlighting public datasets as a crucial resource. Section 3.1

"Using Existing Datasets

" explicitly states

"The easiest way to collect data is to use publicly available datasets... Many websites share public datasets

such as Kaggle

the UCI ML Repository

and the KDD Cup." This confirms that using such repositories is a primary and efficient method. (DOI: https://doi.org/10.1109/TKDE.2019.2946162)

Q: 4

An organization concerned about the ethical and responsible use of a newly developed AI product should consider implementing:

Options

Discussion

Robin Jul 7, 2026 11:02 am

Option C, Saw similar wording on a practice set and it matched the exam answer key.

Aaron S. Jul 22, 2026 5:57 am

Saw something like this in a mock test before and C was the expected answer there too. The accountability model covers governance and ethics, not just transparency. Pretty sure that's what they want in this context.

Grace W. Jul 18, 2026 11:57 pm

Yeah C is the pick for this, covers the full ethical governance angle.

Robin Jul 14, 2026 3:39 pm

Option C here. Accountability model is what covers full governance and responsible use, not just transparency or documentation like A. I think that's what the question wants, but open if someone has a different angle.

Olivia C. Jul 21, 2026 3:37 am

Remember seeing a nearly identical question in a practice bank, and C was the correct choice there too. Ethical and responsible use points right to an accountability model over the others.

Arjun Jul 7, 2026 3:05 am

Pretty sure it's C. Saw something like this in an exam dump, and accountability models cover all the governance and oversight for ethical AI use. The others (like model cards) help with transparency but don't handle the full responsibility angle. If someone thinks differently, open to discussion.

ArjunU Jul 9, 2026 1:47 pm

Probably C. An accountability model sets up clear roles and oversight, so you actually have a structure for ethical AI use. The other options help, but without defined accountability you can't enforce responsible practices. Pretty sure that's what exam writers want here.

SanjayD Jul 16, 2026 10:34 pm

A? Saw a similar question in an old exam report where model cards came up for responsible AI docs.

Vikram X. Jul 25, 2026 3:04 pm

I don't think D fits here, even though 'security by design' sounds tempting. The question is really about overall governance and ethics, which is what C (accountability model) covers. Security alone doesn’t cover responsible use as broadly. Pretty sure C is the safest pick unless I’m missing something.

Logan P. Jul 23, 2026 6:02 am

C - accountability model is what they look for in these ethics questions. Saw similar wording in recent real exam threads, and C matches best for broad responsible use (not just AI transparency). Pretty confident here but let me know if someone read it differently.

Be respectful. No spam.

Correct Answer:

Explanation

An accountability model establishes a clear governance structure, defining roles, responsibilities, and oversight mechanisms for AI systems. This framework is fundamental to ensuring that an AI product is used ethically and responsibly throughout its lifecycle. It addresses the organization's concerns by creating clear lines of authority for managing risks related to fairness, bias, transparency, and societal impact. This model provides the foundation upon which specific tools and practices, such as model cards and security protocols, are implemented and enforced.

Why Incorrect

A. Model cards are a tool for transparency, which is a component of a responsible AI strategy, but they do not constitute the entire governance framework needed for accountability.

B. Vendor monitoring is relevant only when using third-party AI systems and does not address the organization's internal responsibility for a product it has developed.

D. Security by design is a critical practice focused on protecting the AI system from threats, but it does not cover the full spectrum of ethical considerations like fairness and bias.

References

1. ISACA

"Artificial Intelligence: An Audit and Assurance Framework

" 2023. Page 14

under the "AI Governance Framework" section

states

"A governance framework establishes accountability

roles and responsibilities

and decision rights... It also provides a structure for oversight to ensure that AI systems are aligned with the organization’s ethical principles."

2. National Institute of Standards and Technology (NIST)

"AI Risk Management Framework (AI RMF 1.0)

" January 2023. Section 3.1

"The GOVERN Function

" emphasizes that this function is central to responsible AI. It states

"Governance processes should be in place to ensure accountability for AI risks and their management... This includes assigning roles and responsibilities for all stages of the AI lifecycle."

3. Mittelstadt

B. (2019). "Principles alone cannot guarantee ethical AI." Nature Machine Intelligence

1(11)

501-507. This academic paper argues that high-level ethical principles are insufficient without practical implementation through "mechanisms of accountability" and robust governance structures to translate principles into practice. (DOI: https://doi.org/10.1038/s42256-019-0114-4)

Q: 5

Which of the following key risk indicators (KRIs) is MOST relevant when evaluating the effectiveness of an organization’s AI risk management program?

Options

Discussion

Chris M. Jul 12, 2026 4:12 pm

Option C. Saw a similar question in exam reports, and compliance percentage is usually what they're looking for as a real KRI.

Liam Jul 23, 2026 8:38 pm

I saw a similar question on a practice set and went with B.

Ava C. Jul 11, 2026 8:15 pm

C is the pick here. Measures like compliance rates are classic KRIs in ISACA guides for showing how well an AI risk management process is doing, not just how much AI is used. Pretty sure practice questions back this up, but let me know if you see differently.

Grace Z. Jul 17, 2026 11:22 am

B or C. If they're only looking at breadth of AI in critical systems, B stands out for risk exposure since more coverage usually means higher risk surface. But depends if they're judging exposure vs effectiveness-I think B fits better for coverage metrics honestly.

Chloe R. Jul 11, 2026 6:03 pm

NoahU Jul 10, 2026 4:34 pm

C , compliance % is the clearest measure of risk management effectiveness here.

Vikram Jul 17, 2026 10:21 am

C , compliance percentage ties right to how effective those controls and policies actually are. Not totally sure though.

Kevin Jul 22, 2026 5:56 am

I don't think it's B, C is better. B just shows how much AI you have in key systems but doesn't say anything about risk management performance. C directly reflects whether controls and policies are actually being followed. Training requests (D) is a classic distractor for these. Agree?

Jack L. Jul 10, 2026 5:52 am

Yeah, C makes the most sense to me. Compliance rates directly show if controls and policies for AI risk are being followed, unlike A or D which don't really track effectiveness. B is more about exposure than management. Pretty sure C is what exam wants here, but open to other takes if I missed something.

Ben Y. Jul 7, 2026 9:59 pm

Is there a scenario where A would ever be more relevant as a KRI here? Just curious if exam reports mention that.

Be respectful. No spam.

Correct Answer:

Explanation

A Key Risk Indicator (KRI) for an AI risk management program should measure the program's effectiveness in governing AI initiatives and ensuring they adhere to established policies and controls. The "Percentage of AI projects in compliance" is the most direct measure of this effectiveness. It quantifies how well the organization's AI activities are following the prescribed risk management framework, including mandatory assessments, controls, and documentation. A high compliance rate indicates a successful and effective program, while a low rate serves as an early warning that the program is not being implemented properly, increasing overall AI-related risk.

Why Incorrect

A. Number of AI models deployed into production: This is a volume or activity metric. It indicates the scale of AI adoption and potential risk exposure but does not measure the effectiveness of the program managing that risk.

B. Percentage of critical business systems with AI components: This metric measures the organization's inherent risk or attack surface related to AI. It identifies where risk management is crucial but does not evaluate how well it is being performed.

D. Number of AI-related training requests submitted: This is an ambiguous indicator. It could signify a positive culture of risk awareness or, conversely, a lack of adequate foundational training, but it does not directly measure the program's control effectiveness.

References

1. NIST AI Risk Management Framework (AI RMF 1.0): The "Measure" function of the framework is dedicated to tracking risk management effectiveness. It states

"Measurement enables learning from experience and improves the design

development

deployment

and use of AI systems." A compliance metric directly aligns with this goal of evaluating and improving risk management practices. (Source: NIST AI 100-1

January 2023

Section 4.4

"Measure

" Page 21).

2. ISACA

"COBIT 2019 Framework: Governance and Management Objectives": While not AI-specific

COBIT provides the foundational principles for IT governance that ISACA applies to new domains. The management objective APO12

"Manage Risk

" includes example metrics like "Percent of enterprise risk and compliance assessments performed on time." The "Percentage of AI projects in compliance" is a direct application of this established principle to the AI domain

measuring adherence to the defined risk management process. (Source: COBIT 2019 Framework

APO12

Page 113).

3. Thelen

B. D.

& Mikalef

P. (2023). "Artificial Intelligence Governance: A Review and Synthesis of the Literature." Academic literature on AI governance emphasizes the need for "mechanisms for monitoring and enforcement" to ensure compliance with internal policies and external regulations. A KRI measuring the percentage of projects in compliance is a primary tool for such monitoring and enforcement

directly reflecting the governance program's effectiveness. (This is a representative academic concept; specific DOI would vary

but the principle is standard in AI governance literature).

Q: 6

When integrating AI for innovation, which of the following can BEST help an organization manage security risk?

Options

Discussion

QuietReviewer2186 Jul 10, 2026 6:27 am

B , saw a similar question on a mock and they pushed for outside perspectives for risk. Might be off here.

Sara W. Jul 16, 2026 8:04 pm

D is the better choice, seen this in practice exams too. Phased rollout lets you adjust controls as new risks pop up, instead of guessing everything up front. Official guide and sample scenarios cover this approach a lot. Pretty confident but open to other arguments.

Jamie A. Jul 10, 2026 5:36 pm

Honestly, I'm not sure C really helps manage the security risk specifically, since compliance is more about meeting minimum standards. D (adopting a phased approach) lets you iterate and adapt controls as new AI risks emerge, so I think that's strongest here. A is tempting but doesn't give you practical steps for mitigation. Anyone see a case for B?

LeoT Jul 13, 2026 7:42 pm

C is wrong, B. Getting third-party advice can really help spot blind spots or unseen risks, especially if you're new to AI integration. Sometimes fresh external eyes catch things internal teams miss. I'm pretty sure about B but curious if others disagree.

FriendlyAnalyst6103 Jul 12, 2026 5:08 pm

D imo, phased approach actually lets you catch unknown risks as you scale up AI. B looks tempting but third-party advice isn't as hands-on for security control. Pretty sure D's right but willing to hear why someone might choose B.

Ajay C. Jul 21, 2026 1:22 am

D for sure, phased rollout lets you find security gaps early before full deployment. Risk management is all about controlling the blast radius as you scale. The other options help, but nothing beats a controlled, step-by-step process for new AI stuff. Pretty confident on this but open if someone has a different take.

Kevin Jul 25, 2026 5:21 am

I don’t think C or B are right here. Phased approach (D) is usually the better call for managing risk with new tech like AI, lets you spot issues before going big. Compliance (C) only covers legal/reg chores, not full risk. Anyone think otherwise?

Ravi Jul 18, 2026 6:06 am

C or B. Compliance is always a big deal so C seems solid, but B makes sense for fresh perspectives too.

Morgan D. Jul 21, 2026 5:37 am

Its C, not D. Compliance checks can uncover required controls you’d miss otherwise. D is a trap since it’s not always risk-driven.

QuietLearner508 Jul 16, 2026 4:45 pm

C/D? I like D because rolling out AI step by step helps you test controls and adapt if you find new risks. C is about compliance but doesn't always cover new threats from AI. Pretty sure D edges it here for real-world risk management, but open to other thoughts.

Be respectful. No spam.

Correct Answer:

Explanation

Adopting a phased approach is the most effective strategy for managing the security risks associated with integrating a novel technology like AI. This method allows an organization to introduce AI capabilities incrementally, starting with pilots or limited-scope projects. Each phase serves as a controlled environment to identify, assess, and mitigate emergent security risks before they can impact the entire enterprise. This iterative process facilitates learning, allows for the refinement of security controls and governance policies, and ensures that the complexities and potential vulnerabilities of AI are understood and managed before scaling up. It is a foundational risk management practice for complex technology adoption.

Why Incorrect

A. Re-evaluating the risk appetite is a critical governance activity that sets tolerance levels but does not, by itself, constitute a management strategy for implementation risks.

B. Seeking third-party advice is a valuable, supportive measure for gaining expertise but is not the primary, overarching strategy for managing the integration process.

C. Evaluating compliance requirements is essential for establishing a baseline and avoiding legal penalties but often fails to address novel, technology-specific security risks not yet covered by regulations.

References

1. ISACA

COBIT® 2019 Implementation Guide

2018. Chapter 3

"The Implementation Lifecycle

" details a seven-phase

iterative approach. Phase 4

"Plan

" and Phase 5

"Design

" emphasize planning for incremental implementation. This phased methodology is a core principle for managing risk during the implementation of any significant IT change

including AI. The guide states

"The lifecycle should be iterative

meaning that the enterprise can cycle through the phases as needed." This directly supports a phased approach for managing complex integrations.

2. ISACA

"Auditing Artificial Intelligence

" White Paper

2021. Page 14 discusses the "AI Journey" and the importance of starting with pilot projects. It states

"The AI journey is a marathon

not a sprint... It is important to start small with a few pilot projects to learn and build momentum." This recommendation to "start small" is the essence of a phased approach to manage the risks and complexities of AI adoption.

3. Hiekkanen

et al. "This Time It’s Different? A Review of the AI Governance Literature." Proceedings of the 53rd Hawaii International Conference on System Sciences

2020. (https://doi.org/10.24251/HICSS.2020.638). The paper discusses AI governance frameworks

which often implicitly or explicitly recommend iterative and adaptive strategies for AI adoption. The concept of "sandboxing" and pilot programs is highlighted as a key mechanism for exploring AI's potential while containing risks

which is a form of a phased approach.

Q: 7

Which area of intellectual property law presents the GREATEST challenge in determining copyright protection for AI-generated content?

Options

Discussion

Vikram L. Jul 11, 2026 3:35 am

Makes sense that B is the main issue here. Copyright law wants a human author, and figuring out who (if anyone) actually owns AI-generated stuff is still a big gray area. Licensing (D) matters too, but ownership needs to be solved first. Pretty sure that's what most study resources highlight, let me know if you see it differently.

Morgan A. Jul 22, 2026 4:29 pm

I don’t think it’s D, ownership (B) is the bigger challenge since copyright law still wants a human author.

Anita H. Jul 6, 2026 10:07 am

B is the right pick here.

Ishaan H. Jul 15, 2026 1:16 am

B tbh

Quinn S. Jul 12, 2026 9:02 pm

Pretty sure it's B. Copyright law still needs a human author, so ownership is the biggest issue with AI-generated content.

Layla I. Jul 26, 2026 9:47 am

B tbh, saw a similar question on a practice set. The real problem is figuring out who owns the work when AI generates it, since copyright law expects a human author. Not 100% but that's what most guides highlight.

Noah H. Jul 23, 2026 9:02 pm

I get why some might go for D, but it's B here. The big hurdle is figuring out who the actual owner is when AI creates something, since copyright needs a human author. Pretty sure that's what exam guides say too, but open to other opinions if I've missed something.

Avery S. Jul 18, 2026 8:13 am

I don’t think it’s B. D, since setting up licensing frameworks seems like the main challenge for AI content copyright to me.

Nora U. Jul 18, 2026 4:33 am

I don’t think D is right, B makes more sense since the big issue is about who owns what AI creates.

Reese X. Jul 11, 2026 5:44 am

A is wrong, B. Biggest headache is about who actually owns the AI-generated content, since copyright usually needs a human author. The other options come after you figure that out. Open to other takes but that's how I've seen it.

Be respectful. No spam.

Correct Answer:

Explanation

The greatest challenge in applying copyright law to AI-generated content is determining rightful ownership. Traditional copyright frameworks, such as the U.S. Copyright Act, are predicated on the principle of "human authorship." Because an AI is not a legal person, it cannot be considered an "author" in the legal sense. This creates a fundamental conflict: if there is no human author, it is unclear who, if anyone, owns the copyright. The debate involves whether ownership should fall to the user providing the prompt, the AI developer, the owner of the computing resources, or if the work should enter the public domain. This foundational issue of authorship and ownership must be resolved before other issues, like licensing, can be effectively addressed.

Why Incorrect

A. Enforcing trademark rights is concerned with the branding of the AI system itself (e.g., its name or logo), not the copyright of the content it generates.

C. Protecting trade secrets applies to the AI's underlying algorithms, models, and training data, which are distinct from the copyright status of the AI's output.

D. Establishing licensing frameworks is a secondary issue that depends entirely on first resolving the primary challenge of who holds the ownership rights to be licensed.

References

1. U.S. Copyright Office. (2023). Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence. Federal Register

88(51)

16190-16194.

Page 16192

Section III.A: The guidance states

"the Office will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author." This directly highlights that the lack of human authorship is the central barrier to establishing copyright

and thus ownership.

2. World Intellectual Property Organization (WIPO). (2020

May). WIPO Conversation on Intellectual Property (IP) and Artificial Intelligence (AI): Revised Issues Paper on Intellectual Property Policy. WIPO/IP/AI/2/GE/20.

Page 11

Issue 11: Authorship and Ownership: The paper explicitly poses the core question: "Should copyright be attributed to original literary and artistic works that are autonomously generated by AI? If so

who should be considered the author or owner of the copyright...?" This frames the ownership question as the primary policy challenge for international IP bodies.

3. Stanford University Human-Centered AI (HAI). (2023

September 12). Generative AI and the Future of Work.

In the section on "Intellectual Property

" the report discusses the legal ambiguity surrounding AI-generated works

stating

"Current U.S. copyright law protects only works with human authors

leaving the ownership of purely AI-generated content in a gray area." This reinforces that ownership is the central unresolved issue.

Q: 8

When documenting information about machine learning (ML) models, which of the following artifacts BEST helps enhance stakeholder trust?

Options

Discussion

Taylor C. Jul 13, 2026 7:50 pm

C . Model cards are designed for transparency and cover intended use, risks, and performance in a way non-technical folks can digest. B is great for internal controls but doesn't address broad trust. Pretty sure C is what they want, but open to counterpoints.

Ava O. Jul 24, 2026 12:43 pm

C or B? Had something like this in a mock and C was correct because model cards explain limitations and intent in plain language, not just technical details. That helps everyone understand what the model does. Pretty sure it's C here, anyone disagree?

Robin R. Jul 12, 2026 3:44 am

Why is it always "model card" for these trust questions? I've seen similar on other practice sets, and C is the only option that actually targets both technical and non-technical stakeholders, not just internal devs.

Meera O. Jul 25, 2026 11:48 pm

Maybe B since I had something like this in a mock and data quality controls were highlighted. They help ensure input data is reliable which should boost trust, right? Not 100% on this one.

Kevin Y. Jul 8, 2026 11:55 am

Option C Model cards are specifically meant for sharing model details with a wide range of stakeholders. Hyperparameters and data controls matter but don't give a full, understandable picture. Not 100% if everyone uses model cards in production but that's the intent. Disagree?

Ethan H. Jul 8, 2026 8:00 pm

B tbh, since data quality controls are often highlighted in official guides as key for stakeholder confidence on the exam.

Ishaan Jul 15, 2026 5:57 pm

I don’t think it’s B. C provides a structured summary for everyone, not just technical teams, so stakeholders actually understand the model. B is important but doesn’t show transparency the same way. Agree?

AaronM Jul 11, 2026 12:50 am

C is the way to go here. Model cards give a clear overview to all stakeholders, not just technical folks, so they're best for building trust. B is helpful but not as broad. Pretty confident, but let me know if you see it differently.

Nathan Jul 7, 2026 4:55 pm

Ivy C. Jul 7, 2026 9:29 am

Be respectful. No spam.

Correct Answer:

Explanation

A model card is a standardized documentation artifact designed to increase transparency and accountability for machine learning models. It provides a structured summary of a model's intended uses, performance metrics (often disaggregated across different groups), limitations, ethical considerations, and the data used for training and evaluation. By presenting this crucial information in a concise and accessible format, model cards enable various stakeholders—including developers, policymakers, and end-users—to understand the model's capabilities and risks. This transparency is a cornerstone for building trust in AI systems, as it demonstrates due diligence and provides a basis for informed decision-making.

Why Incorrect

A. Hyperparameters: These are low-level technical settings for the training algorithm. They are not meaningful to most stakeholders and do not describe the model's real-world performance or impact.

B. Data quality controls: While essential for building a reliable model, these are processes and metrics related to the input data, not a comprehensive summary document about the finished model itself.

D. Model prototyping: This is an early, experimental phase in the model development lifecycle. A prototype is not a formal documentation artifact for a deployed model and lacks the rigorous evaluation needed for trust.

References

1. ISACA

Artificial Intelligence Security and Management (AAISM) Study Guide

1st Edition

2024. Chapter 3

"AI Model Development and Training

" emphasizes the need for comprehensive documentation to ensure transparency and accountability. It identifies model cards as a key tool for documenting model details

performance

and limitations to communicate effectively with stakeholders.

2. Mitchell

Zaldivar

et al. (2019). Model Cards for Model Reporting. Proceedings of the Conference on Fairness

Accountability

and Transparency

pp. 220-229. This foundational academic paper introduces model cards as a framework to "encourage transparent model reporting" and provide stakeholders with essential information to "better understand the models." (DOI: https://doi.org/10.1145/3287560.3287596)

3. Stanford University

Center for Research on Foundation Models (CRFM). (2023). Transparency Section. The courseware and publications from Stanford's AI programs

such as those from the CRFM

consistently highlight the role of artifacts like model cards in achieving AI transparency and building trust. They are presented as a best practice for responsible AI development and deployment.

Q: 9

An attacker crafts inputs to a large language model (LLM) to exploit output integrity controls. Which of the following types of attacks is this an example of?

Options

Discussion

Amelia G. Jul 25, 2026 3:30 am

Option A. Had something like this in a mock, fits prompt injection best here.

EthanC Jul 18, 2026 1:01 am

C/D? I keep seeing similar questions on practice exams, but the official guide really highlights prompt injection for LLM attacks targeting output controls. Anyone else leaning A after going through the latest sample tests?

Drew M. Jul 8, 2026 12:30 pm

Nah, I don’t think it’s D here. A is the better fit since prompt injection is all about crafting inputs that mess with how the LLM processes output, targeting the model’s interpretation itself. D (evasion) usually means avoiding detection or filtering, not directly manipulating output logic. B (jailbreaking) is tempting but is more about bypassing full system restrictions. Pretty sure ISACA wants A for this scenario. Agree?

Olivia I. Jul 19, 2026 11:08 am

B or D here, since jailbreaking is often about bypassing controls too. A feels more direct but the wording threw me off.

Nora Z. Jul 17, 2026 9:13 am

A or B? Crafting inputs to mess with output integrity is really prompt injection (A), but if those crafted prompts are used to push the model past checks, some call that jailbreaking (B). The key is whether the question wants the attack method or intent-here it's more about manipulating via the prompt, so A fits best. Still, ISACA sometimes blurs these lines in their wording. Anyone see a question where B was preferred in this context?

Arjun Y. Jul 26, 2026 9:47 am

My pick: it's A, since official study materials and practice tests always point to prompt injection for LLM input manipulation scenarios. B sometimes comes up but only when the attack is about totally bypassing restrictions, not just changing outputs. Anyone checked the official guide chapter on LLM attacks?

Mason K. Jul 11, 2026 8:42 am

A tbh, since prompt injection is about feeding crafted prompts to mess with what the LLM outputs. Jailbreaking (B) usually means bypassing all kinds of restrictions, not just output integrity controls. Saw similar logic used in practice sets.

Piya S. Jul 16, 2026 8:41 pm

A Had similar phrasing in official practice tests and the ISACA study guide. This one's straight prompt injection, not evasion.

Zoe Z. Jul 7, 2026 7:57 pm

Its D since evasion is about getting past security controls-seems like that's what they're doing by exploiting output controls. Pretty sure that's the angle here, but could be missing something.

Casey F. Jul 17, 2026 8:56 pm

Be respectful. No spam.

Correct Answer:

Explanation

The scenario describes an attacker crafting specific inputs to bypass an LLM's built-in restrictions and manipulate its output. This technique is known as prompt injection. The attacker injects malicious instructions within the prompt, causing the model to ignore its original system-level instructions and follow the attacker's commands instead. This directly exploits the model's interpretation of input to compromise its output integrity, making it the most accurate description of the attack.

Why Incorrect

B. Jailbreaking is a specific goal or outcome of a prompt injection attack, aimed at bypassing safety and ethical filters, rather than the attack technique itself.

C. Remote code execution involves executing arbitrary code on the server, a more severe and distinct attack that is not inherently part of manipulating LLM text output.

D. Evasion is an adversarial attack that typically causes a model (e.g., a classifier) to make an incorrect prediction, which is different from overriding an LLM's core instructions.

References

1. OWASP Foundation. (2023). OWASP Top 10 for Large Language Model Applications. In the "LLM01: Prompt Injection" section

the vulnerability is defined as: "Prompt injection vulnerabilities allow a malicious user to manipulate the output of a Large Language Model (LLM) through crafted inputs... This can lead to data exfiltration

unauthorized access

or other security breaches."

2. National Institute of Standards and Technology (NIST). (2023). Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations (NIST.AI.100-2 E2023). Section 3.2.1

"Data/Input Poisoning

" describes how attackers can manipulate inputs. Prompt injection is cited as a prime example for LLMs where "an attacker can carefully craft a prompt to an LLM to have it ignore previous instructions."

3. Greshake

Abdelnabi

Mishra

Endres

Holz

& Fritz

M. (2023). Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection. arXiv preprint arXiv:2302.12173. Section 2

"Background

" defines prompt injection as an attack where an "adversary controls a part of a prompt... to manipulate the LLM's behavior." (https://doi.org/10.48550/arXiv.2302.12173)

Q: 10

Which of the following is MOST important to consider when validating a third-party AI tool?

Options

Discussion

Adam Q. Jul 18, 2026 1:26 pm

Gotta go with B here. The ability to audit the vendor is crucial for real validation, not just relying on their word or compliance docs. Pretty sure that's what ISACA is looking for.

Ethan M. Jul 15, 2026 9:08 am

B tbh

Aaron Jul 22, 2026 2:14 am

B . Had something like this in a mock exam and right to audit was flagged as most critical for third-party AI tools. It's all about being able to independently verify how the AI operates and handles your data, not just relying on what the vendor claims. C sounds good but isn't enough for full validation. Others agree?

Mia F. Jul 12, 2026 12:45 pm

Official study guides always push for audit rights (B) when you're dealing with third-party tools. Practice questions I've seen echo this too.

Alex Jul 8, 2026 1:38 am

I get why some argue C but it's gotta be B for long-term control.

Quinn Q. Jul 16, 2026 1:46 pm

I don’t think it’s C. B is stronger since relying just on certifications can be a trap, but audit rights actually let you dig in. Saw similar focus in practice sets.

Arjun Q. Jul 24, 2026 4:28 pm

Had something like this in a mock and B was the right call. Audit rights let you actually verify controls and compliance, not just rely on vendor promises or shiny certs. Pretty sure that's what they're testing for here, but open to other views.

Emma H. Jul 23, 2026 7:32 pm

C vs B, I feel like industry certs (C) carry a lot of weight too but maybe that's the trap here.

Leo Jul 11, 2026 4:26 pm

Its B since right to audit means you can actually check what the third-party tool does, not just trust their paperwork. C is tempting but industry certs are surface level. I think audit rights is what most exam prep points to, agree?

ChloeT Jul 17, 2026 11:25 am

Official practice tests and guides both highlight right to audit (B) as a must for validating third-party AI tools.

Be respectful. No spam.

Correct Answer:

Explanation

The right to audit is the most critical consideration because it provides the organization with the contractual authority to independently verify and validate the third-party AI tool's performance, security controls, data handling processes, and compliance with legal and ethical standards. Given the often opaque nature of AI models, this right is a fundamental governance mechanism for ongoing assurance. It allows the organization to move beyond relying solely on vendor attestations and certifications, enabling direct assessment against specific internal requirements and risk tolerances, which is the core of validation.

Why Incorrect

A. Terms and conditions: While essential, this is a broad legal document. The right to audit is a specific, crucial clause within the terms and conditions that directly enables validation.

C. Industry analysis and certifications: These are valuable for initial due diligence but are often generic, point-in-time assessments that may not cover the organization's specific use case or risk profile.

D. Roundtable testing: This is a specific, often informal, testing technique. It is only one small component of a comprehensive validation strategy, not the most important overarching consideration.

References

1. ISACA

Auditing Artificial Intelligence

2021. Chapter 5

"Auditing AI Governance and Risk Management

" emphasizes the importance of managing third-party AI risks. It states that contracts with AI service providers should include "clauses for independent security assessments and the right to audit to ensure that the organization’s data are adequately protected and that the AI solution is performing as expected." (Page 63).

2. ISACA

Artificial Intelligence: An Audit/Assurance Framework

2023. In the "GOVERN" function

section 3.4

"Third-Party Management

" discusses the need to "Establish and monitor third-party relationships to ensure that AI-related risks are managed." The right to audit is a primary mechanism for such monitoring and for gaining assurance over the third party's controls and processes.

3. NIST

AI Risk Management Framework (AI RMF 1.0)

January 2023. The GOVERN function (Section 4.1

GOVERN 3) discusses policies and procedures for third-party AI systems. It highlights the need for organizations to have mechanisms to understand and manage risks from externally sourced AI

for which audit and assessment rights are a prerequisite for effective implementation. (Page 21).

Q: 11

Which of the following should be the PRIMARY consideration for an organization concerned about liabilities associated with unforeseen behavior from agentic AI systems?

Options

Discussion

CarefulMentor1874 Jul 12, 2026 10:09 pm

Option C is correct. D looks tempting but accountability model directly addresses liability concerns for agentic AI, not just risk appetite.

Mason O. Jul 7, 2026 3:15 pm

Option C seen on similar exam topics, makes sense since accountability model directly addresses how liability is assigned for AI decisions.

RowanQ Jul 16, 2026 12:41 am

I don’t think D is right here. The question specifically wants the main factor for liability, so C (accountability model) fits best. D is more general risk tolerance, but doesn’t assign responsibility if something goes wrong. Anyone else agree?

Grace E. Jul 23, 2026 10:51 am

D. not C

Robin M. Jul 5, 2026 9:33 pm

Its C, though I've seen similar scenarios in practice tests where D looked possible. Might help to check the official guide on governance for agentic AI systems if you're still split.

Luna D. Jul 14, 2026 6:20 am

C saw a similar question show up in some exam reports and the answer was accountability model.

Rowan Jul 23, 2026 8:19 pm

C , liability always ties back to how you assign accountability for decisions in these systems.

QuietLead2295 Jul 24, 2026 1:03 am

Wouldn't accountability (C) always come before setting risk thresholds (D) if the question is about legal liability, not just risk management?

Chris E. Jul 21, 2026 5:21 pm

C/D? C (accountability model) makes more sense for "liabilities" since legal frameworks always go back to who is responsible if an agentic AI goes off-script. D is tempting because risk thresholds matter, but if you're talking legal and ethical exposure, clearly defined accountability is the real foundation. Pretty sure it's C, but happy to hear counterpoints if you think D is stronger for liabilities specifically.

Amelia Jul 8, 2026 6:53 pm

Totally see why people pick D, but liability's main issue is pinning down who’s responsible when AI goes rogue. That’s what an accountability model (C) does. Not 100% sure because wording is a bit vague, but C lines up best with how governance handles this stuff. Disagree?

Be respectful. No spam.

Correct Answer:

Explanation

The primary consideration for managing liabilities from agentic AI's unforeseen behavior is establishing a robust accountability model. This model defines and assigns responsibility for the AI's decisions and outcomes throughout its lifecycle. When an autonomous system acts in an unexpected way, a clear accountability framework is essential for determining legal and ethical responsibility, guiding incident response, and addressing potential harm. It provides the foundational governance structure for managing the specific risk of liability.

Why Incorrect

A. Model dependencies are a technical aspect of an AI system's architecture; while important for operational risk, they do not primarily address the legal framework for liability.

B. Using approved base models is a risk mitigation control, but it does not eliminate the possibility of unforeseen behavior or define who is liable when it occurs.

D. An acceptable risk level is a strategic business decision that sets the threshold for risk tolerance, but it is not the mechanism for managing liability when that threshold is breached.

References

1. ISACA

Auditing Artificial Intelligence

2023: Chapter 2

"AI Governance

" emphasizes that a key principle of AI governance is accountability. It states

"Accountability for AI systems should be established

with clear roles and responsibilities for the development

deployment

and ongoing monitoring of AI systems." This directly links accountability models to the management of AI systems and their outcomes.

2. National Institute of Standards and Technology (NIST)

AI Risk Management Framework (AI RMF 1.0)

January 2023: The "Govern" function is foundational to the framework. Section 3.1

GOVERN-1

highlights the need to establish "policies

processes

procedures

and practices for AI risk management

including roles

responsibilities

authorities

and lines of communication." This establishes an accountability structure as a primary step in managing AI risk

including liabilities.

3. Stanford University Human-Centered AI (HAI)

An Action Plan for Advancing Trustworthy AI

2023: The report repeatedly stresses the importance of accountability mechanisms. On page 10

it calls for "clear accountability mechanisms to ensure that there is recourse and redress when AI systems cause harm

" identifying this as a critical component for building trustworthy AI and managing its societal impact

including liability.

Q: 12

Which of the following is the MOST effective way to mitigate the risk of deepfake attacks?

Options

Discussion

Jordan Y. Jul 20, 2026 12:22 am

Option C, not D. Provenance checks beat LLM detection since D just spots fakes but C actually blocks them.

Chris N. Jul 6, 2026 11:25 am

C had this exact scenario in a practice set and provenance checks were correct there too.

SharpConsultant1639 Jul 15, 2026 5:05 am

C imo, provenance checks go deeper than just detection and actually prevent spread. D is tempting if you only think about identifying fakes but that's not full prevention. Saw this logic on similar exam questions.

Ethan Jul 11, 2026 11:36 pm

Aaron Jul 19, 2026 3:14 pm

Honestly D seems like the way to go here. LLMs are improving at spotting fake content, so automating detection feels pretty effective.

SeasonedCandidate421 Jul 16, 2026 8:02 pm

C or D, but I see C showing up as best practice in the ISACA official guide and practice tests. Labs that focus on media provenance checks really help drill this too. Anyone else think the exam likes C for this kind of scenario?

Quinn Jul 17, 2026 11:55 am

C vs D? C looks more solid to me since verifying provenance stops deepfakes at the source, not just after the fact. D is flashy but LLMs aren't foolproof for detection. Open if anyone thinks I'm missing something.

Arjun T. Jul 9, 2026 7:07 pm

C is what I'd pick. Validating the data's origin directly addresses authenticity, which really stops deepfakes before they even enter the system. Saw similar logic in a practice set, but open if someone thinks otherwise.

Nathan Z. Jul 16, 2026 3:04 am

Not A, C. Had a similar question in an exam report and provenance checks came up as best mitigation.

Nathan X. Jul 8, 2026 10:21 pm

C tbh. Provenance checks (C) actually cut off deepfakes before they spread since you’re validating the original source. D sounds good, but LLMs can miss stuff and are only detection-not prevention. Seen similar in practice sets and C always lines up as most effective.

Be respectful. No spam.

Correct Answer:

Explanation

The most effective and proactive strategy to mitigate the risk of deepfake attacks is to validate the provenance of the data source. Data provenance involves establishing and verifying the origin, history, and chain of custody of a digital asset (e.g., video, audio, image). By using techniques like cryptographic signatures, digital watermarking, or blockchain-based ledgers, an organization can confirm that the media has not been tampered with and originates from a trusted source. This approach addresses the core issue of authenticity, making it a more fundamental and reliable control than reactive detection methods or fallible human judgment.

Why Incorrect

A. Relying on human judgment for oversight: Humans are notoriously unreliable at detecting sophisticated deepfakes, as these are specifically designed to deceive human perception. This makes human judgment a weak primary control.

B. Limiting employee access to AI tools: This is an internal control that does not prevent externally generated deepfake attacks, which are a common threat vector (e.g., a phishing email with a deepfaked video).

D. Using a general-purpose large language model (LLM) to detect fraud: General-purpose LLMs are designed for text-based tasks and are not the appropriate tool for analyzing the complex visual or audio artifacts characteristic of deepfakes.

References

1. ISACA

Artificial Intelligence Audit Toolkit

2023. The toolkit emphasizes the importance of data integrity and governance as foundational controls for AI systems. In the section on AI-specific risks

it discusses the need for controls that ensure the authenticity and integrity of data inputs

which directly relates to validating provenance to counter threats like deepfakes. (Reference to Data Integrity and Input Validation controls).

2. Guera

& Delp

E. J. (2018). "Deepfake Video Detection Using Recurrent Neural Networks." 2018 15th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). While focusing on detection

the paper's context highlights the challenges and the "arms race" between generation and detection

implicitly supporting the need for more fundamental solutions like provenance. The limitations of detection underscore the superiority of proactive authentication. (DOI: 10.1109/AVSS.2018.8639163).

3. MIT Computer Science and Artificial Intelligence Laboratory (CSAIL)

"Detecting Photoshopped and Fake Images." Research from MIT and other institutions consistently points to the difficulty of detection and the value of source verification. Course materials on media forensics often present provenance as a more robust solution than post-facto detection. (Reference to general principles taught in courses like MIT's 6.869 Advances in Computer Vision).

Q: 13

In the context of generative AI, which of the following would be the MOST likely goal of penetration testing during a red-teaming exercise?

Options

Discussion

Layla T. Jul 8, 2026 5:59 am

Makes sense to go with A. Red teaming pen tests for generative AI are really about those unexpected outputs using adversarial inputs.

Luke C. Jul 7, 2026 7:23 am

Option A similar topic came up on an official practice test. Exam guide focuses on adversarial inputs for pen testing.

Nathan Jul 13, 2026 12:05 pm

Nah, I don't think it's B. The real point of pen testing with generative AI is to see if adversarial prompts can get unexpected outputs, so A fits best here. B is more model robustness, not red-teaming. Pretty sure that's what exam guides say.

Morgan J. Jul 24, 2026 4:16 pm

Feels like B, seen this phrased similarly in some practice sets and official guide reviews.

NoahT Jul 10, 2026 8:32 pm

Seen similar in practice sets, pretty sure it matches A. Official guide has good breakdowns on red-teaming AI too.

Casey Jul 9, 2026 4:15 pm

A tbh, saw a similar theme in some exam reports. Pen testing for gen AI usually targets unexpected outputs with crafted prompts, fits red team objectives.

Meera N. Jul 19, 2026 4:32 pm

Probably B, but does the question mean "most likely" from a security or reliability perspective? That could flip my pick.

SteadyDev3235 Jul 7, 2026 9:23 am

B tbh

Leo L. Jul 19, 2026 2:56 am

A gets my pick. In red-teaming generative AI, you're usually crafting adversarial prompts to force the model into unexpected or unsafe outputs, not just stressing it or scrambling everything. Pretty sure that's what pen testers focus on for this context, but correct me if you see it differently.

Avery Jul 14, 2026 12:49 am

Option B this time. Stress testing the model’s decision-making gets mentioned in some practice tests as a way to see how it handles edge cases, so I think it fits for pen testing sometimes. Definitely check the official guide though if you’re not sure.

Be respectful. No spam.

Correct Answer:

Explanation

The primary goal of a red-teaming exercise for generative AI, particularly during penetration testing, is to proactively identify and understand potential failures and vulnerabilities. This is achieved by simulating adversarial attacks. Testers craft specific adversarial inputs designed to bypass safety filters or exploit weaknesses in the model's logic, thereby causing it to generate unexpected, harmful, biased, or otherwise unintended outputs. This process directly evaluates the model's resilience against malicious use and helps developers implement more robust safeguards.

Why Incorrect

B. Stress test the model’s decision-making process: Stress testing primarily evaluates system performance and stability under high load or resource constraints, which is distinct from a security-focused red-teaming goal of finding behavioral exploits.

C. Degrade the model’s performance for existing use cases: While an attack might result in performance degradation, the red team's objective is to discover the vulnerability that enables such an attack, not simply to cause the degradation itself.

D. Replace the model’s outputs with entirely random content: This is too specific and narrow. The goal is to elicit a wide range of unintended behaviors, including harmful, biased, or private information, not just to generate random outputs.

References

1. National Institute of Standards and Technology (NIST). (2023). AI Risk Management Framework (AI RMF 1.0).

Reference: Page 31

Section 4.3.3

"Test

Evaluation

Verification

and Validation (TEVV)." The framework states that TEVV includes "red-teaming to search for vulnerabilities and to discover whether the system can be used in unintended ways." This aligns with generating unexpected outputs from adversarial inputs.

2. Perez

et al. (2022). Red Teaming Language Models to Reduce Harms: Methods

Scaling Behaviors

and Lessons Learned. arXiv preprint arXiv:2209.07858.

Reference: Section 2

"Red Teaming Methods." The paper

authored by researchers from Google

DeepMind

and affiliated universities

explicitly defines red teaming as "the practice of adversarially testing a model to find inputs that elicit harmful or otherwise undesirable outputs." This directly supports option A. (DOI not available for this version

but it is a foundational academic paper in the field).

3. MITRE. (2023). MITRE ATLAS™ (Adversarial Threat Landscape for Artificial-Intelligence Systems).

Reference: Technique: ML04-T1059

"Prompt Injection." The ATLAS framework details how adversaries "craft malicious inputs to manipulate a language model into generating unintended outputs." This is a core tactic used in red-teaming and penetration testing of generative AI.

Q: 14

A financial institution plans to deploy an AI system to provide credit risk assessments for loan applications. Which of the following should be given the HIGHEST priority in the system’s design to ensure ethical decision-making and prevent bias?

Options

Discussion

Meera P. Jul 17, 2026 11:00 am

Option C D looks tempting since numbers feel fair, but human-in-the-loop (C) actually catches bias AI can miss. Similar practice questions went for C over D for the ethical bit. Open if anyone thinks D is better here.

Logan Jul 25, 2026 4:06 am

Option C, encountered exactly similar question in my exam and it's the right pick.

Zoe Q. Jul 22, 2026 4:12 pm

I don't see how D could be right here. Human-in-the-loop (C) is the key safeguard for ethics and bias, while restricting only to financial metrics misses some real world nuances. C

Anita U. Jul 12, 2026 10:23 am

Maybe C. Encountered exactly similar question in my exam and human-in-the-loop (C) was the best fit for ethics, but I could see D being tempting if it focused only on bias. Anyone else remember this one?

Priya Y. Jul 21, 2026 1:39 pm

I don’t think it’s C. D is more about cutting out bias by sticking to just the numbers so it feels like the safer bet ethically. But I could be missing something since human review also helps. Thoughts?

Jason S. Jul 24, 2026 2:55 pm

Its D

Anita X. Jul 9, 2026 5:08 pm

C or D. If we're talking strict prevention of bias, restricting to just objective financial metrics (D) would lower the chance of human subjectivity creeping in. I get that C covers more ethics but D feels closest if it's about measurable fairness. Agree?

AdamL Jul 26, 2026 3:49 am

C is the one I'd pick here. Human review (C) means a person can spot bias an AI might miss, which covers the ethical aspect way better than just sticking to numbers like D. Pretty sure that's what they're after given "ethical" is called out.

Sean Jul 6, 2026 1:49 am

C , saw something like this on a practice set. Human expert making the final call lets you catch bias and explain decisions. That lines up with AAISM's focus on ethical controls over just relying on objective metrics. Anyone think D could still work?

Mia G. Jul 13, 2026 11:07 am

encountered exactly similar question in my exam. in practice, picked C for the ethical and bias piece.

Be respectful. No spam.

Correct Answer:

Explanation

In high-stakes applications like credit risk assessment, ensuring ethical decision-making is paramount. The "human-in-the-loop" (HITL) approach, where the AI system provides advisory outputs and a human expert makes the final decision, is the highest priority. This design embeds accountability and allows for contextual judgment that an AI model cannot replicate. A human expert can override potentially biased or unfair automated recommendations, providing a critical safeguard against algorithmic discrimination and ensuring that decisions are both explainable and ethically sound. This approach directly addresses the core requirement of preventing bias at the point of decision.

Why Incorrect

A. Regularly updating the model is a standard practice for maintaining performance, but it does not inherently address ethical concerns or bias. New data can perpetuate or even introduce new biases.

B. An appeal mechanism is a crucial reactive control for redress after a decision is made, but it is not a preventative measure integrated into the initial decision-making process itself.

D. Restricting the model to "objective" financial metrics is a flawed strategy, as these metrics can act as proxies for protected characteristics (e.g., zip code correlating with race), leading to systemic bias.

References

1. ISACA

Artificial Intelligence Audit Framework

2023. In the section on AI Governance and Risk Management

the framework emphasizes the need for human oversight

especially for high-risk AI systems. It states

"For high-risk AI systems

human oversight should be in place to allow for human intervention to prevent or minimize harm... The final decision is made by a human." (Domain 1: AI Governance

Section 1.3 Human Oversight and Involvement).

2. High-Level Expert Group on Artificial Intelligence

Ethics Guidelines for Trustworthy AI

European Commission

2019. This foundational document lists "Human agency and oversight" as one of the seven key requirements for trustworthy AI. It specifies that for critical decisions

the system should support human autonomy

with the final determination left to a human. It recommends "human-in-the-loop (HITL)" approaches for high-stakes domains. (Chapter II

Page 14).

3. Barocas

& Selbst

A. D.

"Big Data's Disparate Impact

" California Law Review

Vol. 104

2016. This academic paper explains how seemingly neutral or "objective" criteria can function as proxies for protected classes

leading to discriminatory outcomes. It demonstrates why simply restricting data inputs (as suggested in option D) is insufficient to prevent bias. (Section II.A. The Problem of Proxies

pp. 679-685). DOI: https://doi.org/10.15779/Z38G655.

Q: 15

An organization recently introduced a generative AI chatbot that can interact with users and answer their queries. Which of the following would BEST mitigate hallucination risk identified by the risk team?

Options

Discussion

Logan Jul 25, 2026 9:40 pm

D . Fine-tuning is always what ISACA likes for AI hallucination cases. Saw this on similar practice exams and official guide too.

Ethan K. Jul 24, 2026 7:21 am

Option D fine-tuning is what actually cuts down hallucination in these AI chatbot cases.

Cameron K. Jul 15, 2026 1:19 pm

JamieC Jul 21, 2026 1:32 am

Model testing sounds like it'd catch most obvious errors before rollout. A

Liam D. Jul 22, 2026 12:46 am

Saw this pop up in recent exam reports. Anyone else see D picked for hallucination risk mitigation?

Nora R. Jul 14, 2026 3:56 pm

A or B both seem reasonable. Model testing (A) should catch hallucinations before users see them, and larger training sets (B) can help with model accuracy overall. Pretty sure one of these would work. Disagree?

Ethan Jul 6, 2026 8:55 am

Its D fine-tuning directly targets hallucinations for orgs using generative AI.

FocusedArchitect8124 Jul 17, 2026 12:00 am

Just to be clear, is the question asking for the BEST way to mitigate hallucination risk after deployment, or during model development? If they're focused on initial deployment, I'd probably lean D, but if it's about ongoing risk management or general controls, option A might fit better.

DrewR Jul 20, 2026 8:58 pm

D is the move here. Fine-tuning adapts the AI to your domain, which official guides and practice tests always point to as best for reducing hallucinations. If you want more, check the ISACA official study guide or sample exam questions. Pretty sure this matches exam logic, but open if you see it differently.

CarefulSec1839 Jul 23, 2026 1:37 am

Wouldn’t D (fine-tuning) be the strongest here, since that actually adjusts the model’s outputs to your domain and data? Model testing (A) helps spot issues but doesn’t fix the root cause. Curious if anyone thinks A is even close for "BEST" in real deployment scenarios?

Be respectful. No spam.

Correct Answer:

Explanation

Fine-tuning is a process that adapts a pre-trained foundational model to a specific, narrower domain by continuing the training process on a smaller, curated, and high-quality dataset relevant to the organization. This technique "grounds" the model in the organization's specific context and factual data. By constraining the model's knowledge to this trusted information, it directly reduces the likelihood of it generating fabricated or nonsensical information (hallucinations) when answering user queries. It is the most direct technical control listed for mitigating this specific risk.

Why Incorrect

A. Performing model testing and validation is a detective control that identifies the presence and frequency of hallucinations but does not, by itself, mitigate or prevent them.

B. Training the foundational model on large data sets is what creates the base model. This broad, uncurated training is often a source of hallucinations, not a specific mitigation for a deployed application.

C. Ensuring model developers have been trained in AI risk is a procedural and preventative control. While important for governance, it does not directly alter the model's output or reduce hallucinations.

References

1. ISACA

Generative AI: A Practical Audit Approach

2023: In the section "Key Risks and Controls for Generative AI

" the risk of "Inaccurate or Fabricated Information (Hallucinations)" is discussed. A key control mentioned is to "Fine-tune with domain-specific data...to help ground the model in a specific context

reducing the likelihood of generating irrelevant or fabricated information." (Page 11

Table 1).

2. Stanford University

Center for Research on Foundation Models (CRFM)

On the Opportunities and Risks of Foundation Models

2021: The report explains that fine-tuning (referred to as adaptation) is a primary method for specializing a foundation model for downstream tasks. This specialization improves performance and alignment with desired behaviors

such as factuality. (Section 4.2

"Adaptation

" pp. 85-87).

3. Zhang

et al.

A Survey on Hallucination in Large Language Models: Principles

Taxonomy

and Mitigation

ACM Computing Surveys

2024: This peer-reviewed academic survey categorizes mitigation strategies. It places fine-tuning under "Model-based" mitigation techniques

describing it as a method to align the model's knowledge with factual data sources

thereby directly reducing knowledge-based hallucinations. (Section 5.2.2

"Supervised Fine-tuning"). DOI: https://doi.org/10.1145/3626428

Question 1 of 20

Isaca AAISM

Premium Access Includes

FLASH OFFER

avail 10% DISCOUNT on YOUR PURCHASE

Free AAISM Practice Test Questions and Answers (2026) | Cert Empire Practice Questions

Premium Access Includes

FLASH OFFER

avail 10% DISCOUNT on YOUR PURCHASE