CISSP Practice Test

Answer: C
Option C is correct: Clean agents covered by NFPA 2001, such as FK-5-1-12 (Novec 1230), HFC-227ea (FM-200) and inert-gas blends, extinguish a fire without leaving conductive or corrosive residue and without wetting energized equipment, which is why they are the standard choice for rooms full of live server racks. Option A ignores that dry-chemical powder is corrosive and abrasive and contaminates every board it settles on, so it is avoided around electronics in favor of clean-agent portables. Option B overstates what pre-action buys: the dry pipe only guards against accidental discharge, and when a real fire trips the system it still floods energized equipment with water. Option D ignores that CO2 at extinguishing concentration is lethal; NFPA 12 requires pre-discharge alarms and evacuation, and it is not the preferred agent for normally occupied rooms.

Answer: A
Bell-LaPadula is a confidentiality-focused model (no read up, no write down) designed to prevent unauthorized disclosure, while Biba is an integrity-focused model (no read down, no write up) designed to prevent unauthorized or improper modification of data. Reversing these goals (option B) describes the models backward. Neither model addresses availability as its primary goal. The two models use opposite rule structures and are not interchangeable, since they protect different security properties.

Answer: A
A TPM is a dedicated, tamper-resistant hardware chip that securely generates and stores cryptographic keys and can measure and attest to the integrity of the boot process, matching the scenario exactly. A software-based key file lacks hardware tamper resistance and can be copied or exfiltrated more easily. A generic USB drive provides no cryptographic attestation or tamper protection. An unencrypted configuration file offers no protection for keys at all and is the opposite of a secure solution.

Answer: A
Option A is correct: RSA is orders of magnitude slower than AES and can only encrypt a payload smaller than its modulus (about 190 bytes for RSA-2048 with OAEP), so it is reserved for transporting the short session key; AES then encrypts the bulk message quickly, and the key travels safely because only Party B's private key can unwrap it. Option B gets the division of labor wrong; RSA key transport supplies key distribution, not integrity, and neither layer as described provides integrity, which needs a MAC, an AEAD mode or a signature. Option C claims forward secrecy that RSA key transport cannot deliver: anyone who later obtains Party B's private key can unwrap every recorded session key, which is why TLS 1.3 (RFC 8446) dropped RSA key transport for ephemeral Diffie-Hellman. Option D invents a wear-out property; with proper padding, RSA does not weaken with the number of encryptions, and the reason for the split is performance and payload size, not key fatigue.

Answer: B
Option B is correct: Clark and Wilson (1987) model commercial integrity with constrained data items that may be changed only by certified transformation procedures, integrity verification procedures that confirm the data is in a valid state, and separation of duties among the people allowed to run those procedures, expressed as subject-program-object access triples. Option A describes the Brewer-Nash (Chinese Wall) model, whose purpose is preventing conflicts of interest, not enforcing well-formed transactions. Option C describes the Bell-LaPadula confidentiality model's state-machine approach; clearances and classifications play no part in Clark-Wilson. Option D describes Biba, the lattice-based integrity model the question contrasts Clark-Wilson against; Clark-Wilson has no integrity labels.

Answer: C
Option C is correct: Under ISO/IEC 15408 (Common Criteria) the seven Evaluation Assurance Levels are assurance packages, from EAL1 functionally tested to EAL7 formally verified design and tested; the level says how thoroughly the claims in the security target were examined, while the security target itself says what the product claims to do. Option A overreads the rating; vulnerability analysis is one assurance family within the evaluation, and even EAL7 certifies rigor of examination against the stated target, not the absence of vulnerabilities. Option B confuses assurance with functionality; the functional requirements come from Part 2 and the security target, and a product can claim few functions at a high EAL or many at a low one. Option D mixes up two programs; cryptographic module validation is FIPS 140-3 under the CMVP, and its security levels 1 to 4 are unrelated to the EAL scale.

Answer: A
Option A is correct: The Trusted Computer System Evaluation Criteria (DoD 5200.28-STD) define the trusted computing base as the totality of protection mechanisms, including hardware, firmware and software, whose combination is responsible for enforcing the security policy; anything inside it can violate the policy if it fails, which is why it is kept as small and verifiable as possible. Option B ties the definition to an evaluation level; a component belongs to the TCB because the policy depends on it, whether or not it has been evaluated, and EAL4 is not a formal-verification level. Option C draws the boundary at the CPU mode; trusted user-mode processes such as the login program and security daemons enforce policy and are part of the TCB. Option D describes an organizational control baseline such as one tailored from NIST SP 800-53, a governance construct, not the system-level protection mechanisms the TCB names.

Answer: C
Option C is correct: Forward secrecy comes from deriving each session key through an ephemeral Diffie-Hellman exchange whose private values are discarded afterward, so a later leak of the server's long-term key gives an attacker nothing with which to recover past session keys; TLS 1.3 (RFC 8446) requires this by removing RSA key transport. Option A describes PKI revocation checking, which neither detects compromise nor protects recorded traffic; a revoked key can still decrypt captures made before revocation. Option B describes the opposite design; deriving session keys from the long-term key is exactly what destroys forward secrecy, and a broken RNG undermines ephemeral exchange. Option D reverses the relationship; per-session isolation is ordinary key separation, and forward secrecy is specifically about the long-term key not exposing earlier sessions.

Answer: D
Option D is correct: A signature under FIPS 186-5 is a hash of the document transformed with the signer's private key; verifying it with the public key proves the content is unaltered and that the private-key holder signed it, which supports non-repudiation, while the document itself stays readable unless separately encrypted. Option A adds availability, which no signature provides; a signature detects alteration but cannot restore or recover a document. Option B credits the signature with confidentiality it does not provide and denies the non-repudiation it exists for; key sharing is a key-management failure, not a property of the mechanism. Option C describes the wrong key; signing uses the signer's private key so anyone can verify, and encrypting a hash with the recipient's public key would prove nothing about the signer.

Answer: D
Option D is correct: PKI answers a compromised certificate with revocation: the CA lists the serial number on a signed certificate revocation list (RFC 5280) and answers OCSP queries about it (RFC 6960), and relying parties that check either source reject the certificate before its natural expiry. Option A treats an end-entity compromise as a root compromise; replacing the trust anchor is a last resort for a compromised CA key, and it would not tell relying parties this one certificate is bad. Option B gets path validation wrong; issuing a replacement leaves the compromised certificate fully valid until it expires or is revoked. Option C is a forward-looking strategy that limits future exposure but cannot inform anyone about a certificate that has already been issued and compromised.

Answer: A
Option A is correct: Side-channel attacks such as Kocher's timing attacks, simple and differential power analysis, electromagnetic and acoustic emanation analysis and cache-timing attacks recover secrets from how the implementation behaves while it computes, which is why constant-time code, blinding and shielding are the countermeasures rather than a different algorithm. Option B describes key-management failures; a side channel extracts a properly generated key by observing the device that uses it. Option C describes protocol and configuration weaknesses; a side channel needs no downgrade because it observes the correctly configured implementation leaking. Option D describes classical cryptanalysis of the algorithm itself; a side channel attacks a mathematically sound cipher through its physical execution.

Answer: D
The secure design principle of failing securely holds that when a component cannot complete its security function the resulting state must still be safe, which for an authorization decision means denying the request. Permitting and queueing converts an availability fault into an authorization bypass for the duration of the outage. Trusting internal network origin is exactly the implicit trust that modern architectures remove, and it would let any compromised internal host transact freely. Alerting on permitted transactions detects the problem without preventing it, and reversal is a compensating business process, not a security control; the error returned should also avoid disclosing internal failure details.

Answer: D
Option D is correct: Protection rings are hierarchical privilege levels enforced by the CPU: Ring 0 code may execute privileged instructions and touch kernel memory, while Ring 3 code must request services through controlled system calls, so a compromised application cannot directly subvert the kernel or other processes. Option A confuses privilege levels with cryptography; rings gate which instructions and memory regions code may use, and no encryption occurs when control passes between them. Option B mistakes a scheduler policy for a security boundary; ring level governs privilege, not CPU time, and the ring model makes no availability guarantee. Option C describes an administrative control over people, not a hardware privilege boundary; rings constrain executing code, and kernel changes are governed by change management, not by the ring model.

Answer: A
This is session fixation, and the OWASP session management guidance names regeneration as the defense: the session identifier must be renewed after any privilege level change, authentication being the principal case, and the application should never accept a session identifier it did not generate. That invalidates whatever the attacker planted. The secure and httponly attributes protect an identifier from interception and script access, which addresses theft rather than fixation. A shorter idle timeout narrows the window but does not remove the flaw. Binding to address and user agent is a supplementary anomaly check that is easily matched by an attacker on the same network.

Answer: A
ABAC evaluates multiple contextual attributes—user, resource, environment (time, device state), and action—through policy rules to make dynamic access decisions, matching the scenario's combination of department, classification, time, and device compliance. DAC lets resource owners grant access at their discretion, without this rule-based attribute evaluation. RBAC assigns permissions based on fixed roles rather than combining multiple dynamic attributes. Pure MAC relies on classification labels and clearance levels alone, not the broader attribute set (time, device compliance) described here.

Answer: B
Option B is correct: NIST SP 800-63B defines multifactor authentication as the use of two or more of the three distinct factor categories: something you know, something you have and something you are; a memorized password plus a possessed OTP token combines knowledge with possession. Option A stacks two inherence factors; two biometrics from one category remain single-factor, and SP 800-63B requires a biometric to be paired with a physical authenticator. Option C stacks two possession factors; stealing the bag that holds both defeats the scheme, which is exactly why a second category is required. Option D mislabels the questions: answers are memorized facts, so both credentials are knowledge factors, and SP 800-63B does not accept knowledge-based questions as an authenticator at all.

Answer: D
Option D is correct: In SAML 2.0 the identity provider is the authentication authority: it verifies the corporate credentials and returns a digitally signed assertion, and the SaaS application, acting as the service provider, trusts that assertion instead of ever seeing a password. Option A swaps protocols and directions: SAML issues assertions about the user to the SaaS application, whereas an OAuth access token authorizes a client to call an API. Option B assigns the service provider's job to the IdP; signature validation and session creation happen at the SaaS side after the assertion arrives. Option C describes account provisioning under SCIM (RFC 7644), a separate lifecycle function; SAML single sign-on does not depend on it and the IdP's role in the flow is authentication.

Answer: B
Option B is correct: Standardized job-function bundles assigned and revoked as a unit are roles, and the administrative benefit the ANSI/INCITS 359 RBAC standard was built for is exactly this: provisioning, de-provisioning and reviewing access per role rather than per individual. Option A names the wrong model and promises too much; ABAC (NIST SP 800-162) decides on attributes and policies rather than pre-built bundles, and attribute-driven decisions still require periodic review of the policies and attribute data. Option C confuses bundles with labels; MAC compares subject clearances against object classifications and is not defined by job functions at all. Option D hands permission decisions to individual owners, which is the per-person sprawl the enterprise is trying to escape, and DAC has no concept of a role bundle.

Answer: C
Option C is correct: Just-in-time privileged access removes standing administrative credentials: a checked-out credential lives only for the approved task, so a stolen or misused one is valid for hours instead of indefinitely, which is the least-privilege objective behind NIST SP 800-53 AC-6 and CIS Control 5. Option A credits the time limit with an accountability benefit that actually comes from individual identities and session recording; a time-limited shared credential is still shared. Option B misreads the checkout as a factor; the temporary credential is still a single secret used in the same channel, and NIST SP 800-63B expects privileged access to keep MFA. Option D describes an approval-workflow property some PAM tools add; the time limit on the credential itself does not create separation of duties.

Answer: D
This is privilege creep, and the provisioning lifecycle answers it. NIST SP 800-53 account management requires reviewing accounts for compliance with account management requirements at a defined frequency, and the CISSP outline pairs account access review with provisioning and deprovisioning across onboarding, transfers and offboarding. Removing entitlements as part of the transfer process prevents accumulation, while periodic review catches what the process misses. Dual approval improves the quality of new grants but does nothing about ones already held. Password lifetime is unrelated to entitlement scope. Logging shows what was used and leaves the unnecessary access in place.

Answer: A
Mandatory access control means access decisions are made by the system according to a policy expressed through security labels on objects and clearances for subjects, and subjects, including the creator of an object, cannot override that policy or pass access along. The refusal despite the owner's wishes is the defining characteristic. Under discretionary access control the owner holds exactly that discretion, which is what the scenario denies. Role-based access control assigns permissions through job functions but does not by itself prevent an owner from sharing. Risk-based access control adjusts decisions on contextual signals, none of which appear here.

Answer: B
NIST SP 800-53 requires testing backup information at an organization-defined frequency to verify media reliability and information integrity, and the CISSP outline lists backup verification data among the security process data an organization collects. Only an actual restore shows that the media is readable, the data is complete and consistent, and the documented procedure works within the target recovery time. Confirming retention alignment is a paper comparison. Offsite transfer records show copies moved, not that they can be read. Job success dashboards report what the backup software believed it did, which is the very claim the auditor is refusing to accept unverified.

Answer: B
Black-box testing simulates an external attacker who has no prior internal knowledge of the target system's architecture, source code, or credentials, matching the scenario exactly. White-box testing provides the tester full internal knowledge, including architecture and source code, which contradicts the 'no prior system knowledge' condition. Gray-box testing gives partial knowledge, and explicitly having 'full source code access' would actually make it white-box, not gray-box. Static code analysis examines source code without executing it and is a different technique from a simulated external attack.

Answer: B
A false positive occurs when a scanner flags a vulnerability that does not actually pose exploitable risk in the real deployed environment, exactly as described when manual verification shows the code path is unreachable. A true positive would mean the vulnerability is confirmed as genuinely exploitable, which is the opposite of what verification found. A false negative is a missed vulnerability that actually exists but wasn't detected, not a finding that was detected but doesn't apply. A true negative would mean no vulnerability was reported and none exists, which doesn't match this scenario where a finding WAS reported.

Answer: A
Option A is correct: An AICPA SOC 2 Type 2 report opines on both the design and the operating effectiveness of controls across a review period, so the auditor samples evidence of the control running throughout that window, typically six to twelve months. Option D inverts the test: exceptions show where the control failed, and the auditor must evidence the routine operations to conclude it worked. Option B is a Type 1 approach dressed up as Type 2; a point-in-time design test plus an unverified assertion does not evidence operation over the period. Option C samples one instance and extrapolates; Type 2 testing requires items drawn from across the whole period to show the control kept operating.

Answer: A
Option A is correct: A red team exercise pursues a defined objective by whatever path works, usually unannounced beyond a small control group, and the measured outcome is whether the blue team detected and responded; a penetration test instead aims to find and confirm as many exploitable weaknesses as possible within its scope. Option B describes a purple team exercise, which trades the realism of an unannounced campaign for faster detection tuning. Option C describes a security assessment or audit; a red team attacks live systems, which is what exercises detection and response. Option D describes exploitation-focused penetration testing; completeness of findings is the opposite of a goal-oriented campaign that takes one path in.

Answer: A
Option A is correct: Centralized log management (NIST SP 800-92) enables correlation across sources, so a logon on one host, a share access on another and a new service on a third become one visible lateral-movement sequence even though each event looks routine in its own log; forwarding also keeps copies an intruder cannot erase on the compromised host. Option C misdescribes the data and the benefit; events from different systems are distinct records, and retention cost is not the capability that per-system review lacks. Option B promises attribution that aggregation cannot supply; a shared or service account stays shared, and resolving it to a person requires individual accounts or session brokering. Option D misunderstands collection; an event a source never wrote does not exist to be re-requested, and completeness depends on the source's logging configuration.

Answer: D
Option D is correct: Static application security testing analyzes code or compiled binaries without running them, finding flaws early and pointing to the line while producing false positives for unreachable paths; dynamic testing sends inputs to the running application and observes behavior, catching configuration and environment issues static analysis cannot see, which is why NIST SP 800-218 expects both. Option B describes configuration auditing and performance testing; neither is application security testing of the code or the running application. Option A misplaces both; design review is not SAST, and reviewing source code is static, not dynamic, testing. Option C describes software composition analysis and fuzzing; SCA inventories dependencies rather than analyzing the code, and fuzzing is one dynamic technique, not the definition.

Answer: D
The CISSP outline distinguishes administrative, criminal, civil, regulatory and industry-standard investigations, and they differ in who conducts them and in the evidentiary standard required. A matter that could escalate must be handled from the first step to the most demanding standard likely to apply: NIST SP 800-86 stresses acquiring data with documented, repeatable methods and an unbroken chain of custody, because evidence collected loosely cannot later be repaired. Interviewing first tips off the subject and risks destruction of volatile evidence. Referring to law enforcement before any internal assessment forfeits the organization's own fact-finding. Assuming it stays administrative is the error the question tests.

Answer: C
RTO is the targeted duration within which a business process must be restored after a disruption to avoid unacceptable consequences, which matches the 4-hour figure derived during BIA. RPO instead measures acceptable data loss in time (how much data, not downtime, can be lost). MTTR is an operational/technical repair-time metric, not a business-impact-derived target. MTD is the absolute outer limit of tolerable downtime and is typically larger than or equal to the RTO, not the target itself the question describes.

Answer: A
NIST SP 800-161 Rev. 1 treats cybersecurity supply chain risk management as an acquisition discipline: risks such as tampering, counterfeits and implants are addressed by flowing minimum security requirements into agreements, assessing and continuously monitoring suppliers, and demanding provenance evidence such as a software bill of materials. Those controls act while the sourcing decision can still change. Sampling inspection catches only what arrives and only some of it. Insurance transfers cost without reducing the exposure. Firmware scanning is worth doing but happens after the supplier has already been chosen and the component built.

Answer: C
Awareness and training are different instruments. NIST SP 800-50 Rev. 1 describes a learning program that pairs broad awareness with role-based training tied to the duties of specific roles, and that is evaluated with metrics measuring behavior change rather than attendance. Authorization flaws in code are a role-specific skill gap, so more general awareness will not close it; repeating the same content more often changes nothing about its relevance. Discipline alone punishes the symptom without supplying the missing skill. Publishing a standard and collecting attestations records exposure to the document, not competence in applying it.

Answer: B
NIST SP 800-34 Rev. 1 sets out three BIA steps in order: determine mission and business processes and recovery criticality, identify resource requirements, and identify recovery priorities for system resources. Criticality comes first because everything else depends on knowing which processes the system supports and what an outage costs them. Resource identification is the second step and presumes those processes are already known. Developing recovery priorities is explicitly the last step of the BIA process. Choosing a recovery strategy or alternate site happens after the BIA, using its downtime findings as input.

Answer: D
Under 45 CFR 164.402 an impermissible acquisition, access, use or disclosure of protected health information is presumed to be a breach unless the covered entity or business associate demonstrates a low probability that the information has been compromised, based on a risk assessment of at least four factors: the nature and extent of the information including re-identification risk, the unauthorized person involved, whether the information was actually acquired or viewed, and the extent to which the risk has been mitigated. Whether the file was opened is only one of those factors, not the test. A deletion attestation is mitigation evidence feeding the assessment. Demonstrated harm is not the standard, and 164.414 places the burden of proof on the entity.

Answer: B
Option B is correct: The CISSP exam outline effective April 15, 2024 lists eight domains weighted Security and Risk Management 16%, Asset Security 10%, Security Architecture and Engineering 13%, Communication and Network Security 13%, Identity and Access Management 13%, Security Assessment and Testing 12%, Security Operations 13% and Software Development Security 10% (isc2.org/certifications/cissp/cissp-certification-exam-outline). Option A has the right count but the wrong leader; Security Architecture and Engineering is one of four domains at 13%. Option C has the right count but the wrong leader and figure; Identity and Access Management is weighted 13%, below Security and Risk Management. Option D names the right domain but the wrong count and weight; the outline has eight domains and Security and Risk Management carries 16%.

Answer: B
Option B is correct: Under the outline effective April 15, 2024, the CISSP is a three-hour computerized adaptive test of 100 to 150 items with a passing score of 700 out of 1000 scaled points; ISC2's outline page now lists the CAT format for the English, Chinese, German, Japanese and Spanish exams. Option A describes the retired six-hour linear form; the English exam has been adaptive since 2017 and the outline page no longer lists a linear version for any language. Option C has the format right but the scoring wrong; an adaptive exam is scored on a scaled 0 to 1000 range with 700 as the pass mark, not on a percentage of items correct. Option D describes the four-hour, 125-to-175-item CAT format ISC2 used from 2022 until the April 15, 2024 outline replaced it; the item range is now 100 to 150.

Answer: C
Option C is correct: Data sovereignty means the same dataset answers to the access, retention, disclosure and transfer laws of every country whose territory holds a copy, so a multi-region cloud deployment can be bound simultaneously by GDPR Chapter V transfer rules, the US CLOUD Act and local data-localization statutes. Option B gets the doctrine backwards; long-arm statutes extend a court's reach to out-of-jurisdiction parties, and the CLOUD Act shows a foreign-stored copy can be compelled from the provider. Option D is a preservation duty triggered by anticipated litigation; it applies regardless of location and does not create the conflict between jurisdictions' laws. Option A misstates a defunct mechanism; the EU-US Safe Harbor framework was invalidated by the Court of Justice in Schrems I (2015) and never authorized unrestricted holding of anyone's data.

Answer: A
The preamble to the ISC2 Code of Ethics orders the obligations as the safety and welfare of society and the common good, then duty to principals, then duty to each other, and Canon I is to protect society, the common good, necessary public trust and confidence, and the infrastructure. A contract term cannot displace a higher canon, so the member must press the client to remediate and disclose, escalating rather than staying silent. Canon III (service to principals) is real but subordinate here. The Code applies to all professional conduct, not only ISC2 activities, and adherence is a condition of certification. Canon IV does not authorize unilateral publication that could itself endanger users.

Answer: B
ISC2 limits standing by canon: complaints about Canons I and II may come from the general public, complaints about Canon III (diligent and competent service to principals) may be brought only by principals such as employers or those in a contracted relationship, and complaints about Canon IV may be brought only by other certified or licensed professionals who subscribe to a code of ethics. A competitor is none of those for a service-quality claim, and every complainant must also show injury from the conduct. The public-standing rule applies to Canons I and II, not Canon III. Technical knowledge of the work does not create standing, and the Canon IV route covers conduct affecting the profession, not one client engagement.

Answer: B
NIST SP 800-39 lists the available responses as acceptance, avoidance, mitigation, sharing, transfer, or a combination. Insurance shifts financial consequence to an insurer and is the classic transfer (or sharing) response; the threat, the vulnerability and the operational disruption all remain with the buyer. Avoidance means eliminating the activity or technology that creates the risk, which buying a policy does not do. Mitigation reduces likelihood or impact through controls, and a policy changes neither. Acceptance means absorbing the consequence internally without a countervailing arrangement, which is precisely what the policy replaces.

Answer: D
Under the NIST Risk Management Framework the authorizing official is a senior executive who reviews the authorization package and issues an authorization to operate, an authorization to use, a common control authorization, or a denial; accepting residual risk on behalf of the organization is that official's decision, not the security team's or the project's. The security function assesses and advises but does not own the business risk. Blanket refusal ignores that risk can be knowingly accepted when it falls within tolerance and a plan of action and milestones tracks the remaining work. A project manager lacks the authority and the enterprise view the decision requires.

Answer: A
SP 800-37 Rev. 2 defines the authorization to use as the decision employed when an organization chooses to accept the information in an existing authorization package produced by another organization, typically for cloud and shared systems and services. A common control authorization runs the other way: it covers controls a provider offers for inheritance by systems, after the common control provider's own package is reviewed. The framework's interim testing authorizations are not a production authorization based on someone else's package. Reciprocity supports reuse of assessment results, but it is not automatic and does not transfer an authorization boundary by contract.

Answer: B
FIPS 199 expresses a security category as a triple of potential impact values for confidentiality, integrity and availability, and for a system the values are the highest (the high water mark) among the security categories of all information types it processes, stores or transmits. Here that yields moderate, low and high respectively, and FIPS 200 then uses the highest of those to set the system's impact level for baseline selection. Averaging invents a value the standard does not recognize. Volume is irrelevant to impact. Taking the lowest value would leave the most sensitive information type protected below its assessed need.

Answer: C
NIST SP 800-12 Rev. 1 divides policy into three basic types. Program policy creates the organization's information security program, assigns its management, and sets purpose, scope, responsibilities and compliance; it is typically issued by a senior official such as the senior information security officer. Issue-specific policy addresses one topic of current concern, and system-specific policy states the rules for a particular system. A procedure is a step-by-step description of how work is performed and derives from policy rather than establishing it; procedures cannot assign program-level accountability.

Answer: D
Option D is correct: SLE = AV x EF = $500,000 x 0.40 = $200,000; a flood expected once every 4 years gives ARO = 1/4 = 0.25; ALE = SLE x ARO = $200,000 x 0.25 = $50,000. Option A reports the SLE as if it were the ALE; the $200,000 single-loss figure still has to be multiplied by the 0.25 annual rate. Option B inverts the frequency, treating one flood every four years as four floods per year, which multiplies the SLE by 4 instead of by 0.25. Option C multiplies the full asset value by the ARO and skips the exposure factor, pricing a total loss when the scenario destroys only 40% of the asset.

Answer: D
Option D is correct: ISC2 lets a candidate sit the CISSP before meeting the five-year experience requirement; on passing, the candidate is designated an Associate of ISC2 and has up to six years to accumulate the five years of cumulative paid experience in two or more domains (isc2.org/certifications/associate). Option A misstates the waiver: the one-year reduction is granted for a qualifying four-year degree or an approved credential, never for passing the exam itself. Option B adds a retake that does not exist; the passing result stands while the Associate completes the experience inside the six-year window. Option C invents a provisional credential; ISC2 does not award the CISSP, provisionally or otherwise, until the experience is verified and endorsed.

Answer: D
The data owner (often a business-unit executive) has ultimate accountability for a dataset, including assigning its classification level and approving access decisions, because they best understand its business value and sensitivity. The data custodian implements technical controls (backups, access provisioning) on the owner's behalf but does not decide classification. System administrators perform operational tasks per established policy. End users merely consume data according to granted permissions and have no classification authority.

Answer: B
Need to know further restricts access within an already-authorized role so that a subject can only access the specific information required to perform their current task, which is why the accounts-payable employee is denied payroll data despite having general financial-system access. Separation of duties splits a sensitive process across multiple people to prevent fraud, which is a different control from restricting information access scope. 'Mandatory vesting' is not a security access-control concept. Due diligence refers to the reasonable care an organization exercises in managing risk, not an access-restriction mechanism.

Answer: C
NIST SP 800-88 Rev. 2 scales the sanitization method to where the media is going. Clear applies logical techniques to all user-addressable storage locations to defeat simple, non-invasive recovery through the normal user interface, and it typically leaves the media usable, which fits reassignment inside the same organization at the same sensitivity. Destroy is for media leaving control or holding data that must never be recoverable, and it ends the media's usable life, so buying replacements here is wasteful. A format and reinstall is not a recognized sanitization method and leaves data in unallocated space. Degaussing is discussed as a physical purge technique for magnetic media, it is not an approved destroy technique, and it generally renders a drive inoperable rather than reusable.

Answer: B
SP 800-88 Rev. 2 treats cryptographic erase as a purge technique that works by sanitizing the keys, so it can only address data that was encrypted. The publication states the precondition that no sensitive data was previously stored on the media in plaintext form, and warns that cryptographic erase should not be trusted for media that has been backed up or escrowed unless the organization has high confidence in how those keys were stored and managed elsewhere. A recycler's certificate documents a downstream act, not whether the erase was effective. Key strength matters, but the acceptability question here turns on plaintext history and key copies. Nothing restricts cryptographic erase to resale scenarios.

Answer: D
SP 800-88 Rev. 2 notes that for media taking the form of logical or virtual storage such as cloud storage there may be no purge alternative to cryptographic erase, because the physical media are abstracted away and the data owner cannot sanitize them; it advises understanding those options before sensitive data is stored there. Destroying customer-managed keys therefore gives the assurance sought. Degaussing multi-tenant provider hardware is not something a customer can require or verify, and degaussing is not an approved destroy technique. API overwrites cannot reach replicas, snapshots and prior versions the abstraction hides. A delete call removes visibility, not necessarily recoverability.

Answer: A
SP 800-88 Rev. 2 provides a sample certificate of sanitization and lists the details it should record, including manufacturer, model, serial number, media type and source, the sanitization method (clear, purge or destroy) and the specific technique, the tool used including version, the verification method, and the identity, title, date and signature of the person performing verification and validation. That record ties one identified device to one documented act. An asset register shows a replacement was bought, not what happened to the original. A screenshot lacks the device identifiers and the accountable signature. A waste manifest evidences transfer of custody by weight, not sanitization of a particular drive.

Answer: C
NIST SP 800-122 directs organizations to minimize the use, collection and retention of personally identifiable information to what is strictly necessary, observing that the likelihood of harm from a breach is greatly reduced when holdings are minimized; it also has organizations categorize information by PII confidentiality impact level, and identifiers of this kind raise that level. Data that is never collected cannot be breached, mishandled or subpoenaed. Encryption, shortened retention and segregated storage are sound handling controls once a genuine need exists, but each of them accepts the collection of information that has no current purpose and so leaves the underlying exposure in place.

Answer: B
Baselines are starting points that organizations tailor, and when a control cannot be applied as written the recognized path is a compensating control that yields equivalent protection, documented with its rationale and approved by the responsible official. Here that might be surrounding the shared login with logged physical access, session recording and out-of-band identification of the operator. Silently deleting the control from the baseline loses the record of why the requirement is unmet. Marking an unimplemented control as implemented misstates the system's security posture to everyone who relies on it. Deferring to a replacement cycle still leaves the interim period unprotected and undocumented.

Answer: C
Asset retention decisions must account for end of life and end of support, because an unsupported platform stops receiving the fixes the rest of the control set assumes. When the business cannot retire it on time, the defensible response combines technical containment such as network segmentation, restricted administrative access and enhanced monitoring with a time-boxed, formally accepted risk decision and review points. Signature-based endpoint protection does not remediate unpatched operating system vulnerabilities. An abrupt shutdown trades a security risk for a larger operational one without weighing either. Extended support helps but rarely covers every component, so it supplements containment rather than replacing it.

Answer: D
Establishing information and asset handling requirements means the protection travels with the asset. Media protection practice pairs marking that tells every handler the classification with encryption so that loss of a tape does not disclose its contents, and with accountability records that document custody at each transfer. A predictable collection schedule is an operational convenience and arguably tells an observer when the tapes are in transit. Environmental conditions in the vault preserve the media but do nothing about interception on the road. A non-disclosure agreement creates a legal remedy after disclosure rather than preventing it, and the courier should not be able to read the data at all.

Answer: A
The outline requires candidates to determine data security controls across the data states of at rest, in transit and in use. Encryption at rest and in transit leave a gap in the third state: to compute on the values the application must hold them in cleartext in memory, where a compromised process, a memory scrape or an administrator with debug rights can read them. Control options include restricting privileged access, hardened or attested execution environments and tokenizing values so the sensitive form is never processed. Key custody at rest and TLS termination at a proxy are genuine design concerns, but both are described as already addressed. Archives are storage, a subset of data at rest rather than a separate state.

Answer: A
Option A is correct: GDPR Article 4(5) defines pseudonymization as processing that requires separately kept additional information to attribute data to a person, and Recital 26 states that such data remains personal data while truly anonymous information falls outside the Regulation; the practical difficulty is that auxiliary datasets often re-identify records believed anonymous. Option C inverts the transfer rules; pseudonymized data is personal data and needs Chapter V safeguards, while anonymous data needs none. Option B reverses Recital 26; removing direct identifiers while a key exists is pseudonymization and stays in scope, and the survival of some original dataset elsewhere does not make anonymous data personal. Option D keeps anonymous data in scope and invents requirement mappings; anonymized data is outside the regime, and pseudonymization is a safeguard, not a substitute for encryption at rest.

Answer: A
Option A is correct: NIST SP 800-88 (Guidelines for Media Sanitization) defines three levels: Clear resists only simple non-invasive recovery, Purge makes laboratory recovery infeasible, and Destroy additionally leaves the media unusable. Drives holding patient records that will leave the hospital's control for recycling warrant Destroy, which shredding, pulverizing or degaussing to the point of unusability achieves. Option B is a Clear-level technique in SP 800-88; it defeats casual recovery tools but is not the assurance level specified for high-sensitivity media leaving organizational control, and it cannot reach remapped or damaged sectors. Option C removes directory entries and overwrites only the blocks the new OS happens to use; the bulk of the patient data remains on disk and is trivially carved out with forensic tools. Option D encrypts nothing that matters: a quick format leaves the old data in place, and encrypting the empty volume afterward does not touch the remnants; cryptographic erase only works when the data was encrypted from the start.

Answer: B
Option B is correct: The information owner in NIST SP 800-18 and the asset owner in ISO/IEC 27001 Annex A 5.9 is the business role accountable for the data's value, classification and protection requirements, while the custodian is the operational role, usually IT, that runs the backups, permissions and monitoring the owner requires. Option A swaps the roles; classification and sign-off are ownership decisions, and the custodian carries them out rather than making them. Option C reduces ownership to funding and hands access decisions to a user; approving access is an owner responsibility and custodians are operators, not end users. Option D imports privacy-law roles; the individual described is the data subject, and controller and processor are GDPR terms, not the asset-security owner and custodian.

Answer: D
Option D is correct: The PCI Security Standards Council's tokenization guidance describes a token as a surrogate value mapped to the original only through a protected vault, so a stolen token yields nothing without the vault; that is what lets tokenized systems be removed from compliance scope, and it concentrates the risk in the vault. Option A describes hashing, which is neither reversible nor tokenization; the vault exists precisely so the issuer can recover the original when authorized. Option C again describes encryption with a shorter output; a true token has no key relating it to the original, so the obligation is protecting the vault, not a key. Option B describes format-preserving encryption; a reversible, key-based transformation is encryption and inherits its key-management and cryptanalytic exposure.

Answer: A
Option A is correct: Retention limitation is the asset-security principle for sensitive personal and cardholder data: PCI DSS Requirement 3.2.1 limits storage of account data to what legal, regulatory and business purposes require and mandates secure deletion once that period ends, and privacy regimes apply the same storage-limitation rule to Social Security numbers. Option B applies an accounting schedule to data with no accounting purpose and leaves copies in backups and exports, so the deletion is neither minimal nor complete. Option C picks the longest schedule rather than the shortest defensible one, and archiving to tape is still retention, so the data keeps generating breach and discovery exposure. Option D keeps an indefinite copy; encryption reduces exposure but does not satisfy a retention limit, and PCI DSS treats encrypted account data in backups as stored account data.

Answer: C
Stateful firewalls primarily track connection state using Transport-layer (Layer 4) information such as TCP sequence numbers, flags (SYN/ACK/FIN), and port pairs to determine whether a packet belongs to an established session. Layer 1 deals with physical signaling and has no session concept. Layer 2 deals with MAC addressing and frames, not session state. While some stateful and next-gen firewalls incorporate application-layer awareness, classic stateful inspection is defined by its Layer 4 state-tracking, not exclusive Layer 7 operation.

Answer: C
Option C is correct: WPA3-Personal replaces the WPA2 four-way handshake with SAE, which resists offline dictionary attacks on the passphrase, and WPA3-Enterprise uses 802.1X/EAP so each user authenticates with individual credentials; both close the eavesdropping and unauthorized-association weaknesses WEP and WPA2-PSK left open. Option A relies on TKIP, an RC4-based cipher the Wi-Fi Alliance deprecated; rotating a shared key every 90 days still leaves every user holding the same secret and the captured handshake crackable offline. Option B adds two obscurity measures: MAC addresses are sniffed and spoofed in seconds and a hidden SSID is broadcast in every probe response, so neither is an authentication factor. Option D reintroduces WPS PIN, whose eight-digit PIN is brute-forced in hours because the registrar validates each half separately (CERT VU#723755), undoing the protection WPA3 provides.

Answer: C
Option C is correct: The attacker sends small queries to open resolvers with the source address forged to the victim's, so the resolvers reflect large responses at the victim; response sizes many times the query size give the amplification, and the victims never sent the queries they are answering for (CISA alert TA13-088A). Option A is a covert channel that moves data out slowly and quietly; it neither spoofs the victim's address nor tries to saturate the victim's link. Option B describes an integrity attack on resolver caches that changes where names resolve; it does not flood bandwidth, and the scenario shows no altered records. Option D puts the flood in the wrong place and protocol: the resolver is the unwitting reflector, the victims receive UDP DNS responses, and no TCP handshake is being exhausted.

Answer: D
Option D is correct: An 802.1Q VLAN splits one physical switch into separate broadcast domains, and an ACL or firewall at the routed boundary between the guest and finance VLANs decides what, if anything, may cross, so a compromised guest device has no Layer 2 path to the finance servers and only the permitted Layer 3 path. Option A separates wireless names, not networks; unless each SSID maps to its own VLAN with filtering, both land in the same broadcast domain and the guest device can still reach the servers. Option B authenticates whatever plugs into the server ports; it does nothing to control where traffic from an already-connected guest device may go. Option C is defense in depth, not segmentation: it relies on every server's local rules staying correct and on guest addresses staying predictable, and it still lets the guest traffic reach the server's interface.

Answer: C
NIST SP 800-97 recommends implementing 802.1X and EAP for authentication rather than pre-shared keys, because of the administrative burden and security risks of PSK management, and recommends the EAP-TLS method wherever possible since it authenticates with certificates on both sides. That gives per-device credentials that can be revoked individually, which is precisely what the shared passphrase cannot do. Rotating the passphrase repeats the same problem on a schedule. Hiding the network name and filtering hardware addresses are trivially defeated and were never authentication controls. Segmentation plus a VPN is a reasonable defense-in-depth layer but leaves association itself controlled by a widely known secret.

Answer: A
NIST SP 800-121 Rev. 2 advises organizations to use the strongest Bluetooth security mode available for their devices, identifying Security Mode 4, Level 4 for devices supporting Secure Connections because it requires authenticated pairing and FIPS-approved algorithms, with Level 3 for older devices. Security Mode 1 devices never initiate security and the publication says they should never be used, so making them non-discoverable is not a substitute. A fleet-wide fixed passkey reintroduces a shared static secret and defeats authenticated pairing. Choosing service-level security alone is weaker than the link-level protection the current mode provides during establishment.

Answer: B
NIST SP 800-58 addresses both halves of the problem: the Secure Real Time Protocol protects the media stream so conversations cannot be reconstructed by anyone capturing packets, and logical separation of voice from data limits which hosts can reach the telephony infrastructure at all. Quality of service markings address call quality, not security, and putting the call manager among general workstations increases its exposure. Encrypting only signaling leaves the audio itself readable, which is the content most worth protecting. Collapsing handsets onto the data VLAN removes the separation, and port security addresses which device connects rather than what it can reach afterwards.

Answer: C
NIST SP 800-119 devotes substantial attention to dual-stack environments and to the security implications of tunneling mechanisms such as ISATAP, Teredo and configured tunnels, which encapsulate IPv6 inside IPv4 and are transparent to IPv4 infrastructure. A control set written for IPv4 will not see that traffic, giving an unmonitored path in and out. Address length is a tooling detail, not the exposure. IPv6 replaces broadcast with multicast, which changes discovery mechanics but is not the assessor's warning. IPsec support is defined for IPv6 but is not universally mandated in deployment, and it would not describe the uninspected-path problem.

Answer: D
NIST SP 800-81 Rev. 3 states that while DNSSEC helps protect against compromise of DNS communications through integrity protection, it does not provide any confidentiality or privacy protection, and that those capabilities come from other technologies such as DNS over TLS, DNS over HTTPS and DNS over QUIC. Signing adds authenticated signatures to records; it does not encrypt them, and answers remain readable on the wire. Validation is a check on returned data rather than the establishment of an authenticated channel. Zone transfers are protected separately by transaction signatures and access control lists, but DNSSEC does apply to ordinary query responses.

Answer: A
Option A is correct: An 802.1Q VLAN is only as isolating as the switch configuration that tags and filters frames: switch spoofing through auto-negotiated trunks, double tagging across a native VLAN and an administrator's mistake all create paths the design assumed impossible, so VLANs cannot be treated as air gaps. Option B gets the layer wrong; a VLAN is a Layer 2 broadcast domain, and hosts in different VLANs cannot exchange frames without a router or Layer 3 switch. Option C invents a protocol dependency; VLAN tagging operates below IP and confines IPv6 neighbor discovery and multicast to the VLAN just as it confines IPv4. Option D misreads trunking; carrying multiple tagged VLANs is what trunks are for, and the risk lies in misconfigured trunks and native VLANs, not in trunking itself.

Answer: B
Option B is correct: NIST SP 800-207 states the zero trust tenets: no implicit trust based on network location, per-session access decisions, dynamic policy using identity, device posture and context, and continuous monitoring, all aimed at the lateral movement that a perimeter model permits once an attacker is inside. Option C swaps one implicit trust for another; SP 800-207 evaluates user identity, device posture and context per request, and never grants network-wide access on a certificate. Option D confuses confidentiality with authorization; encryption protects traffic in transit but says nothing about whether the requester should reach the resource. Option A restates the perimeter model with a stronger front door; free internal movement after one check is exactly the implicit trust zero trust removes.

Answer: D
Option D is correct: An on-path attacker succeeds by terminating two separate encrypted sessions, one with each victim, so the control that defeats the attack is authentication of the peer: each side validates the other's certificate against a trusted authority (and pins where appropriate), which the impostor cannot satisfy. Option C closes one insertion technique on one LAN; an attacker at a hotspot, an upstream router or a compromised proxy is unaffected, so it does not defeat the attack itself. Option A protects the wrong thing; unauthenticated Diffie-Hellman is the textbook victim of an on-path attacker, who simply runs a separate exchange with each party. Option B assumes the key was negotiated with the right peer; when the attacker is that peer on both legs, the cipher's strength is irrelevant.

Answer: C
Option C is correct: Network access control combines 802.1X authentication of the user or device with a posture assessment of the endpoint, checking patch level, antimalware and agent status and configuration, and steers non-compliant devices to a remediation network, closing the gap of an authorized user on an unhealthy machine. Option A describes authorization policy on the user; it never examines the health of the device that user is bringing onto the network. Option D describes MAC authentication bypass with profiling, the weak fallback for devices that cannot run a supplicant; a spoofed MAC passes, and no posture is checked. Option B describes post-admission intrusion detection; NAC's decision is made before the device is granted access, based on identity and posture.

Answer: B
NIST SP 800-52 Rev. 2 requires that TLS 1.2 configured with approved cipher suites be supported by government TLS servers and clients and requires support for TLS 1.3, and it states that SSL 3.0 is not approved for the protection of federal information because it relies in part on algorithms that are not NIST-approved. Address allowlisting does not repair the cryptography and is trivially spoofable for a client-driven negotiation. Disabling compression and renegotiation mitigates specific attacks but leaves the deprecated protocol versions in use. Prohibiting TLS 1.2 overstates the requirement, since that version remains mandatory to support rather than forbidden.

Answer: C
NIST SP 800-207 splits the policy decision point into a policy engine, which makes and logs the grant or deny decision, and a policy administrator, which establishes or tears down the communication path by instructing the enforcement point. The policy enforcement point is the system responsible for enabling, monitoring and ultimately terminating connections to the resource, so it is where traffic is actually stopped. Reversing the engine and enforcement point misstates the flow, since the engine consults data sources rather than being consulted by the enforcement point. The administrator directs but does not itself sit in the data path. Continuous diagnostics is one of the data sources that feeds the engine.

Answer: B
NIST SP 800-41 Rev. 1 describes a firewall policy that denies all traffic by default and permits only what is explicitly required, and it describes egress filtering as restrictions on traffic leaving the organization's network. Both findings are policy defects, so both need policy fixes. Reordering rules is a performance measure and leaves the permissive rule in place. Logging or detecting outbound abuse identifies exfiltration and command-and-control traffic after it has already left. Deep inspection and authentication improve what a rule does once it matches, but neither closes a default-permit posture or supplies the missing outbound restrictions.

Answer: C
SP 800-41 Rev. 1 distinguishes packet filtering, stateful inspection and application-proxy gatewaying. Only the proxy acts as an intermediary that terminates the client connection and establishes its own connection to the server, which is what allows it to parse, inspect and modify application content. A stateless packet filter decides on header fields and is unconcerned with payload. Stateful inspection adds connection tracking, so it can tell whether a packet belongs to an established session, but it does not reconstruct and rewrite application data. Address translation hides internal addressing and is not a content inspection mechanism at all.

Answer: D
A demilitarized zone is the recognized answer: the publicly reachable server sits on its own protected segment, and firewall policy allows the internet to reach only that segment while permitting only the specific protocol and port it needs into the internal network. A compromise then yields a foothold in a segment that has almost no reach. Port forwarding to an internal host leaves the compromised server on the same segment as the databases. Requiring authentication protects the application but not the network path once the server itself is taken. Placing the server outside the firewall entirely leaves it undefended and still needs a controlled path inward for its data.

Answer: B
Containment is the phase focused on limiting the scope and impact of an incident, such as isolating a compromised server from the network to stop ongoing exfiltration, which is exactly what's described. Preparation involves proactive readiness activities (plans, tooling, training) performed before an incident occurs. Detection and analysis involves identifying and scoping the incident, which has already occurred by the time isolation is triggered. Lessons learned is the post-incident review phase used to improve future response, not an active response action.

Answer: C
A hot site is a fully operational, continuously synchronized duplicate environment capable of near-immediate failover, matching the minutes-long recovery described. A cold site provides only basic infrastructure (power, space, cooling) with no pre-installed systems or data, requiring days to weeks to become operational. A warm site has some equipment and partial data replication but requires additional setup and typically takes hours, not minutes, to fully activate. A mobile trailer with no pre-installed equipment is essentially a portable cold site and would not support minutes-long recovery.

Answer: B
The order of volatility principle dictates collecting the most transient evidence first: CPU registers and cache change constantly and are lost the instant power/process state shifts, followed by RAM (lost on reboot/power-off), then network connection state (changes as sessions open/close), with disk contents being comparatively persistent and collected last. Option A reverses the entire order, starting with the most persistent artifact (disk) and ending with the most volatile (registers), which risks losing the most fragile evidence first. Option C and D scramble the sequence in ways that place persistent disk data ahead of more volatile network or register state, violating OOV and risking loss of transient evidence during collection.

Answer: B
Option B is correct: NIST SP 800-34 Rev. 1 defines MTD as the total time the business can tolerate the process being unavailable and RTO as the target for restoring the system, and it requires the RTO to be shorter than the MTD so that work recovery time, the period needed to restore data and resume processing after systems are back, also fits inside the limit. Option A imports an operational component-repair average; MTTR is not a business-continuity target, and the MTD is set by business impact, not by adding engineering metrics. Option C leaves no margin; a target equal to the limit fails as soon as the recovery runs long or the work recovery period is added, which is the harm the MTD marks. Option D mixes metrics; the RPO measures tolerable data loss in time before the disruption and has nothing to subtract from a downtime limit.

Answer: C
Option C is correct: NIST SP 800-61 and SP 800-86 frame this as a management decision: restoring before the intrusion is understood risks reinstating the vulnerability or the attacker's persistence and destroys evidence, while waiting has business cost, so the usual resolution is to image the affected systems quickly and restore while analysis continues on the images, with business and legal input. Option D hands the decision to a contract; notification clauses require timely notice and cooperation, and no policy obliges the business to stay down until an investigation ends. Option B reduces the decision to one metric; the RTO informs the impact side but ignores evidence preservation and the risk of restoring a still-compromised system. Option A misapplies a collection principle; the order of volatility sequences what is captured first, and imaging lets restoration proceed while analysis, not just collection, continues.

Answer: A
Option A is correct: A SIEM ingests, normalizes and correlates events from authentication systems, firewalls and endpoints in near real time, which is the cross-source correlation NIST SP 800-92 describes for detecting multi-stage attacks that no single tool's console would connect. Option B sees only endpoint telemetry; it cannot join firewall and authentication-server logs to the endpoint alert. Option C inspects packets on one network segment; it has no view of authentication logs or endpoint alerts and does no cross-source correlation. Option D automates response playbooks after something has been detected; it consumes the SIEM's correlated alerts rather than performing the correlation itself.

Answer: D
Option D is correct: SOAR platforms run playbooks that call other tools' APIs to isolate a host, disable an account or block an indicator when a defined alert condition fires, which is what turns a well-understood alert into an automatic action without waiting for an analyst. Option A can isolate its own host but cannot disable a directory account or coordinate actions across other tools, and tamper protection guards the agent, not the workflow. Option B improves detection by scoring anomalous behavior; it generates alerts and does not execute response actions. Option C produce the alert but do not act on it; a correlation rule raises an incident, and executing containment across the EDR, directory and firewall is orchestration.

Answer: D
Option D is correct: Change management frameworks such as ITIL provide an emergency change path authorized by a small emergency change authority, with the risk assessment, backout plan and approval documented even if compressed, and a post-implementation review afterward; NIST SP 800-40 Rev. 4 likewise expects enterprises to plan expedited handling for actively exploited vulnerabilities. Option A leaves 90% of the estate exposed to a known active exploit for weeks; the pilot is sensible only as the first step of the emergency rollout, not as a reason to wait. Option B drops the documented approval and the review; verbal authority with paperwork weeks later gives no backout plan, no traceability and no evidence for auditors. Option C misclassifies the change: standard changes are low-risk, repeatable and pre-authorized, and an unscheduled production patch under active exploitation with no backout plan is none of those.

Answer: B
OWASP input validation guidance prefers allowlist validation, in which input is checked against the type, length, format and range that the application actually expects, over denylisting known bad patterns, because attackers can encode, case-shift or fragment payloads in ways no blocklist anticipates. Maintaining the list on a schedule leaves the application defenseless against anything not yet published. Client-side filtering is a usability feature that any attacker bypasses by calling the endpoint directly. Encryption protects data in transit and does nothing to the content once decrypted, and validation must in any case be paired with context-appropriate output encoding and parameterized queries.

Answer: C
OWASP password storage guidance rejects fast general-purpose hashes including MD5, SHA-1 and SHA-256 for passwords, because their speed lets an attacker with the database make enormous numbers of guesses; the remedy is a purpose-built, tunable password hashing function, with Argon2id preferred and PBKDF2 with a high iteration count named where FIPS-140 validation is needed. Salting a fast hash defeats precomputed tables but not high-speed brute forcing of each entry. Reversible encryption means a key compromise discloses every password in cleartext. Choosing a longer digest changes the output size, not the per-guess cost that makes fast hashes unsuitable.

Answer: A
Option A is correct: The OWASP SQL Injection Prevention Cheat Sheet lists prepared statements with parameterized queries as the primary defense because the query structure is fixed before user data is bound, so input can never be interpreted as SQL regardless of what it contains. Option B is OWASP's secondary defense; it helps, but free-text fields cannot be allow-listed tightly enough to be relied on as the primary barrier. Option C is the approach OWASP marks as strongly discouraged; escaping is database-specific, easy to get wrong and bypassed through encoding and alternate syntax. Option D is a compensating control in front of the flaw; rule sets are evaded with obfuscation and the injection remains in the code.

Answer: D
Option D is correct: NIST SP 800-218 (SSDF practice PW.1) and the Microsoft SDL place threat modeling in design, when data flows and trust boundaries are defined and a threat can still be removed by changing the architecture rather than by patching code later at far higher cost. Option A uses the model only to aim a test; by then the design flaws it would have surfaced are built and can only be confirmed, not prevented cheaply. Option B substitutes one activity for another; threat modeling informs security requirements but cannot replace them and has no architecture to analyze before design begins. Option C arrives after the architecture is fixed, so every design-level mitigation the model identifies now costs a rewrite; a data-flow diagram, not code, is what threat modeling needs.

Answer: D
Option D is correct: Everything served to the browser is public to the user: view-source, developer tools or a proxy exposes the JavaScript verbatim, so hard-coded database credentials (CWE-798) give any visitor direct authenticated access to the database. Option A frames a performance worry and assumes browsers connect to the database directly; the significant concern is disclosure of the secret, not pool exhaustion. Option B muddles caching with transport security; whether or not TLS or a cache is involved, the credentials are readable in the delivered file itself. Option C invokes an unrelated control; the same-origin policy governs script access between origins and has no bearing on secrets printed in a script.

Answer: A
Option A is correct: NIST SP 800-218 (SSDF practice PW.1) calls for threat modeling while designing the software, because a threat found on a data-flow diagram can be removed by changing the architecture, whereas the same threat found later can only be patched, at a cost that rises with every subsequent phase; STRIDE gives the structure for that early analysis. Option B arrives after the architecture is built; a design-level threat visible in review can be patched but no longer designed out cheaply. Option C uses the model only to confirm; by testing time the design is deployed and confirmation does nothing to lower the cost of the fix. Option D waits for the harm; a model built from a real breach is the most expensive way to learn what a design-phase analysis would have shown.

Answer: A
Option A is correct: The OWASP SQL Injection Prevention Cheat Sheet names prepared statements with parameterized queries as the primary defense: the database compiles the query structure first and binds the input afterward as data, so no input value can change the statement, whereas the other measures reduce exposure without removing the flaw. Option B is a secondary defense that breaks legitimate input such as a surname with an apostrophe and cannot cover free-text fields; validation narrows the attack surface but does not fix the concatenation. Option C is defense in depth for the aftermath; a limited account reduces the damage of a successful injection but leaves the injection possible. Option D is OWASP's second option and only safe when the procedure itself avoids dynamic SQL; a procedure that concatenates its arguments is injectable just like the application was.

Answer: B
Option B is correct: Cross-site request forgery abuses ambient authority: because the browser attaches the session cookie to any request bound for the application, a forged form or image request from an attacker's page arrives looking like a legitimate authenticated action, which is why the OWASP CSRF Prevention Cheat Sheet prescribes synchronizer tokens and SameSite cookie attributes. Option A describes a replay attack by an on-path attacker; CSRF originates from a page the victim visits and involves no interception of traffic. Option C describes session fixation, in which the attacker plants a session ID; in CSRF the attacker never learns or sets the victim's session. Option D describes cross-site scripting; CSRF needs no script execution in the application's origin and never reads the cookie, it only causes the browser to send it.

Answer: B
Option B is correct: A library that was clean when selected becomes vulnerable the day a CVE is published against it, so NIST SP 800-218 and OWASP's guidance on vulnerable and outdated components call for an inventory of components (an SBOM) that is continuously checked against vulnerability feeds, which is what lets an organization answer which applications are affected within hours. Option A freezes known-vulnerable versions in place; pinning gives reproducibility, but without monitoring and upgrading, every later disclosure stays unpatched. Option C addresses license risk and provenance; a signed, properly licensed library can still ship a critical vulnerability. Option D cannot inventory dependencies or match them to disclosed vulnerabilities; DAST observes behavior from the outside and misses library flaws the test inputs never trigger.

Answer: D
Option D is correct: NIST SP 800-218 (SSDF practices PO.3 and PW.7 and PW.8) expects security checks to be built into the toolchain and automated, and in a pipeline that deploys on every merge the only control that keeps pace is a gate that runs at pipeline speed and blocks the build rather than reporting after the fact. Option A is a valuable periodic assessment outside the pipeline; it finds flaws months after the automated deploy shipped them. Option B protects the build hosts and the running service; neither control examines the code moving through the pipeline or can stop a bad build. Option C inserts a manual administrative gate that defeats continuous integration and still checks nothing about the code; approval belongs to release governance, not to the pipeline's technical controls.

Answer: B
In the OWASP Top 10:2025 the software supply chain failures category expands the earlier vulnerable and outdated components entry to encompass broader compromise of the chain through which software is built and delivered, including dependencies, build tooling and distribution. Security misconfiguration concerns how systems and applications are configured. Software or data integrity failures is a related but separate category about trusting unverified code and data, particularly in deserialization and update flows. Insecure design addresses flaws present in the design itself, which no amount of correct implementation can remove, and it is also a carried-over category rather than a widened one.

Answer: C
OWASP guidance on error handling is to show users a generic, non-revealing message with a unique identifier they can quote, while the detailed diagnostic information goes to server-side logs; the mishandling of exceptional conditions category in the current Top 10 reflects how often this goes wrong. Verbose errors disclose schema names, query structure, file paths and framework versions that accelerate attacks. Gating detail on a support role puts sensitive internals one authorization flaw away from disclosure. Changing the status code hides errors from monitoring without addressing disclosure. Returning nothing at all leaves users and support with no way to correlate the failure.