IT Cert Ready

AWS

AWS Certified AI Practitioner

Exam code: AIF-C01

Start practicing

About this certification

AWS Certified AI Practitioner is a AWS certification. Exam code AIF-C01. IT Cert Ready offers 1,101 practice questions with domain quizzes, full mock exams, and detailed explanations. Practice by official domain with 4 blueprint-aligned domains.

Official exam domain weights

Verified blueprint weights from the official exam outline.

  • Domain 1 · Fundamentals of AI and ML · 20%
  • Domain 2 · Fundamentals of Generative AI · 24%
  • Domain 3 · Applications of Foundation Models · 28%
  • Domain 4 · Guidelines for Responsible AI · 28%

Sample AWS Certified AI Practitioner practice questions

Preview realistic multiple-choice items from the IT Cert Ready question bank. Expand each answer for the correct choice and explanation—the same format used in domain quizzes and mock exams.

  1. Sample question 1 · Guidelines for Responsible AI (28%)

    Hargrove & Pellini LLP, a 120-attorney regional law firm, recently deployed an AI-powered contract review assistant that processes uploaded client documents. A paralegal reports that the tool surfaced a summary containing confidential settlement terms from a different client's matter — terms that were never part of the uploaded document. The IT security team suspects the AI model may be retaining and cross-contaminating data across user sessions. The firm's managing partner demands an immediate response, but the vendor's SLA allows up to 72 hours for incident acknowledgment. Outside counsel for the firm's malpractice insurer is asking whether the firm's acceptable use policy for AI tools addresses data isolation requirements. Given the active data privacy incident and the escalating business and legal exposure, which action should the firm's AI governance lead prioritize FIRST to contain risk and satisfy responsible AI obligations?

    • A. Notify the malpractice insurer immediately and wait for their guidance before taking any technical action, to avoid voiding coverage.
    • B. Suspend the AI tool's access to client matter data and initiate an incident review to determine whether the acceptable use policy requires data isolation controls for AI systems handling privileged information.
    • C. Update the firm's acceptable use policy to retroactively include data isolation language, then resume normal AI tool operations while the vendor investigates.
    • D. Request that the vendor provide a root cause analysis within the 72-hour SLA window and take no internal action until that report is received.
    View answer & explanation

    Correct answer: B. Suspend the AI tool's access to client matter data and initiate an incident review to determine whether the acceptable use policy requires data isolation controls for AI systems handling privileged information.

    [Domain 4 - Guidelines for Responsible AI, Subtopic 4 - Responsible AI - Area 1 - Data privacy] Core Concept: Identify security, compliance, and risk considerations. requires applying data privacy in the context of the scenario, consistent with the official (ISC)² exam outline. Correct Answer (B): Suspend the AI tool's access to client matter data and initiate an incident review to determine whether the acceptable use policy requires data isolation controls for AI systems handling privileged information. Why the other options are incorrect: Option A: Waiting for insurer guidance before taking technical action allows the data privacy incident to continue and may worsen legal exposure; containment is a firm's independent obligation regardless of insurance coordination. Option C: Retroactively updating the acceptable use policy does not address the active incident and resuming operations without root cause analysis exposes additional client matters to the same cross-contamination risk. Option D: Deferring entirely to the vendor's 72-hour SLA ignores the firm's independent duty to protect client confidentiality and delays containment actions that are within the firm's own control.

  2. Sample question 2 · Guidelines for Responsible AI (28%)

    Vantara Broadcasting, a mid-sized regional media company, recently deployed an AI-powered phishing detection tool to flag suspicious emails targeting its newsroom staff. The tool was trained on historical phishing datasets and is now used to automatically quarantine messages before employees see them. During a quarterly review, the security team discovers that the model disproportionately flags emails from freelance journalists in certain geographic regions as social engineering attempts, even when those messages are legitimate. Reporters in those regions are missing time-sensitive story tips. The compliance officer warns that the pattern may constitute discriminatory filtering under the organization's acceptable use policy and applicable data privacy regulations. Which action BEST addresses the responsible AI risk identified in this scenario while maintaining the security function of the phishing detection tool?

    • A. Disable the phishing detection tool entirely until a new model can be trained from scratch, eliminating any risk of biased filtering during the interim period.
    • B. Audit the training data and model outputs for demographic or geographic bias, implement guardrails that route flagged messages from affected sender profiles to human review, and update the acceptable use policy to reflect the corrective controls.
    • C. Expand the quarantine scope to all external email senders so that the filtering criteria are applied uniformly, removing any appearance of geographic discrimination.
    • D. Notify affected freelancers to use an alternative communication channel and document the workaround as a compensating control in the risk register.
    View answer & explanation

    Correct answer: B. Audit the training data and model outputs for demographic or geographic bias, implement guardrails that route flagged messages from affected sender profiles to human review, and update the acceptable use policy to reflect the corrective controls.

    [Domain 4 - Guidelines for Responsible AI, Subtopic 4 - Responsible AI - Area 1 - Data privacy] Core Concept: Identify security, compliance, and risk considerations. requires applying data privacy in the context of the scenario, consistent with the official (ISC)² exam outline. Correct Answer (B): Audit the training data and model outputs for demographic or geographic bias, implement guardrails that route flagged messages from affected sender profiles to human review, and update the acceptable use policy to reflect the corrective controls. Why the other options are incorrect: Option A: Disabling the tool entirely removes the security control without addressing the bias root cause, creating an unacceptable security gap and failing to apply responsible AI remediation principles. Option C: Expanding quarantine to all external senders amplifies the operational disruption and does not resolve the underlying model bias; uniform application of a flawed rule does not constitute a responsible AI control. Option D: Redirecting affected freelancers to an alternative channel treats the symptom rather than the cause, does not remediate the biased model, and may itself raise data privacy concerns by routing communications outside established secure channels.

  3. Sample question 3 · Guidelines for Responsible AI (28%)

    Orindal Healthcare Network operates 12 regional clinics and recently deployed an AI-assisted endpoint detection and response (EDR) platform managed by a third-party security operations provider, Clarvex SecOps. The platform uses a machine learning model to prioritize endpoint alerts and recommend remediation actions. During a compliance audit, the internal audit team finds that the Clarvex SecOps model was trained on a dataset that significantly overrepresents large urban hospital environments. Orindal's rural clinic endpoints generate alert patterns that differ substantially from the training data, causing the model to consistently deprioritize legitimate threats at rural sites. The audit team flags this as a risk to patient data security. Which responsible AI consideration most accurately describes the root cause of the alert deprioritization issue at Orindal Healthcare Network's rural clinics?

    • A. Insufficient guardrails preventing the EDR platform from generating alerts above a configurable severity threshold.
    • B. A violation of the vendor's acceptable use policy due to Orindal deploying the platform in an environment outside its licensed scope.
    • C. Model bias resulting from a training dataset that does not adequately represent the operational environment in which the model is being applied.
    • D. A data privacy breach caused by the EDR platform transmitting rural clinic endpoint data to Clarvex SecOps without patient consent.
    View answer & explanation

    Correct answer: C. Model bias resulting from a training dataset that does not adequately represent the operational environment in which the model is being applied.

    [Domain 4 - Guidelines for Responsible AI, Subtopic 4 - Responsible AI - Area 1 - Data privacy] Core Concept: Identify security, compliance, and risk considerations. requires applying data privacy in the context of the scenario, consistent with the official (ISC)² exam outline. Correct Answer (C): Model bias resulting from a training dataset that does not adequately represent the operational environment in which the model is being applied. Why the other options are incorrect: Option A: Guardrails are controls that constrain model outputs or behaviors, but the scenario describes a training data representativeness problem, not a missing output constraint; adding a severity threshold guardrail would not correct the underlying bias. Option B: There is no indication in the scenario that Orindal is operating outside a licensed scope; the issue is model performance degradation due to distributional mismatch, not a policy violation. Option D: The scenario does not describe unauthorized data transmission or a consent failure; the risk identified by the audit team is about biased model outputs, not a data privacy breach event.

  4. Sample question 4 · Guidelines for Responsible AI (28%)

    Aurelius Energy Systems, a privately held offshore wind energy operator headquartered in the Netherlands with operations across the North Sea, has integrated a generative AI assistant into its security operations center (SOC) to help analysts triage network attack indicators and draft incident response runbooks. A penetration test commissioned by the company's cyber risk insurer reveals that the AI assistant has no content guardrails configured, allowing analysts to inadvertently prompt the model to generate detailed technical instructions for exploiting vulnerabilities in industrial control systems (ICS). The insurer flags this as a material control gap that could violate the company's cyber insurance acceptable use obligations and expose the utility to regulatory sanctions under the EU's NIS2 Directive. To close the identified control gap and satisfy both the insurer's acceptable use requirements and NIS2 compliance obligations, which action should Aurelius Energy Systems prioritize?

    • A. Configure content guardrails on the generative AI assistant to detect and block prompts or outputs that request or produce exploit code, attack methodologies, or ICS vulnerability details, and document the guardrail policy in the utility's acceptable use framework.
    • B. Migrate the generative AI assistant to a private cloud deployment so that all model interactions remain within the corporate network perimeter, eliminating external data privacy exposure.
    • C. Require all SOC analysts to complete annual cybersecurity awareness training that includes a module on responsible AI use before they are permitted to access the AI assistant.
    • D. Disable the AI assistant's ability to access external threat intelligence feeds, reducing the volume of sensitive data the model can reference when generating runbooks.
    View answer & explanation

    Correct answer: A. Configure content guardrails on the generative AI assistant to detect and block prompts or outputs that request or produce exploit code, attack methodologies, or ICS vulnerability details, and document the guardrail policy in the utility's acceptable use framework.

    [Domain 4 - Guidelines for Responsible AI, Subtopic 4 - Responsible AI - Area 1 - Data privacy] Core Concept: Identify security, compliance, and risk considerations. requires applying data privacy in the context of the scenario, consistent with the official (ISC)² exam outline. Correct Answer (A): Configure content guardrails on the generative AI assistant to detect and block prompts or outputs that request or produce exploit code, attack methodologies, or ICS vulnerability details, and document the guardrail policy in the utility's acceptable use framework. Why the other options are incorrect: Option B: Moving the assistant to a private cloud changes the deployment topology and may address some data residency concerns, but it does not prevent the model from generating harmful technical content; the control gap identified by the insurer is about output risk, not network perimeter exposure. Option C: Awareness training is a valuable administrative control but relies on human judgment to prevent misuse; it does not technically enforce acceptable use boundaries and would not satisfy an insurer or regulator seeking demonstrable automated controls against harmful output generation. Option D: Disabling external threat intelligence feeds reduces the model's information sources but does not prevent it from generating exploit guidance based on its pre-trained knowledge; the core vulnerability — absence of output filtering — remains unmitigated.

  5. Sample question 5 · Guidelines for Responsible AI (28%)

    Voltara Grid Services, a mid-sized regional energy utility in the Pacific Northwest, has deployed an AI-based anomaly detection system to monitor its operational technology (OT) network for indicators of cyberattacks on substation control systems. During a compliance audit, the auditor discovers that the AI model was trained on raw SCADA telemetry logs that include employee login credentials, device access timestamps, and shift-schedule metadata tied to named individuals. The utility's data governance team has not documented a data minimization review, and no privacy impact assessment was completed before training began. The auditor must recommend a corrective action that addresses the most critical responsible AI compliance gap. Which corrective action most directly addresses the responsible AI compliance gap identified during the audit?

    • A. Retrain the AI model using only anonymized or de-identified telemetry features, and complete a retroactive privacy impact assessment documenting what personal data was used and how it will be remediated.
    • B. Expand the model's training dataset to include additional SCADA logs from partner utilities so that individual employee patterns are diluted across a larger population.
    • C. Apply differential privacy noise to the model's output predictions so that adversaries cannot reverse-engineer individual employee schedules from the anomaly scores.
    • D. Restrict access to the AI system's dashboard to senior security analysts only, and log all queries to satisfy audit evidence requirements.
    View answer & explanation

    Correct answer: A. Retrain the AI model using only anonymized or de-identified telemetry features, and complete a retroactive privacy impact assessment documenting what personal data was used and how it will be remediated.

    [Domain 4 - Guidelines for Responsible AI, Subtopic 4 - Responsible AI - Area 1 - Data privacy] Core Concept: Identify security, compliance, and risk considerations. requires applying data privacy in the context of the scenario, consistent with the official (ISC)² exam outline. Correct Answer (A): Retrain the AI model using only anonymized or de-identified telemetry features, and complete a retroactive privacy impact assessment documenting what personal data was used and how it will be remediated. Why the other options are incorrect: Option B: Adding more data from partner utilities does not eliminate the personal data already embedded in the training set; it compounds the privacy risk by potentially introducing additional individuals' information without consent or minimization review. Option C: Applying differential privacy to model outputs addresses inference-time privacy but does not remediate the fact that personal data was ingested into the training pipeline without a privacy impact assessment — the root compliance gap remains unresolved. Option D: Restricting dashboard access is an access control measure, not a data privacy control; it does not address the unauthorized use of personal data during model training and does not satisfy the audit finding related to data minimization or impact assessment documentation.

AWS Certified AI Practitioner exam prep FAQ

What is the passing score for the AWS Certified AI Practitioner exam?
Passing scores for AWS Certified AI Practitioner are set by AWS and can change between exam versions. Check the latest official exam outline on the AWS website for the current passing standard.
How many questions are on the AWS Certified AI Practitioner certification exam?
Official AWS Certified AI Practitioner exam length and item types are defined by AWS. Review the current exam blueprint for the latest question counts and timing.
How do domain quizzes help with AWS Certified AI Practitioner exam prep?
Domain quizzes on IT Cert Ready let you practice AWS Certified AI Practitioner one blueprint domain at a time across 4 official exam domains. You get immediate feedback, detailed explanations, and can revisit weak areas before taking a full timed mock exam.
How many AWS Certified AI Practitioner practice questions are on IT Cert Ready?
IT Cert Ready includes 1,101 published AWS Certified AI Practitioner practice questions with domain quizzes, full-length mock exams, and answer explanations mapped to the official outline.