SCS-C03 practice questions and answers
All 65 questions from Full Practice Test 1 for AWS Certified Security - Specialty, with the correct answer and a full explanation for each — including why the other options are wrong. Free to read, no signup.
What this set covers
Questions are weighted to match the official SCS-C03 exam guide. The real exam is 65 (50 scored) questions in 170 minutes with a pass mark of 750 / 1000.
- Detection10 q · 16%
- Incident Response9 q · 14%
- Infrastructure Security12 q · 18%
- Identity and Access Management13 q · 20%
- Data Protection12 q · 18%
- Security Foundations and Governance9 q · 14%
Which two practices reduce the risk associated with the AWS account root user? (Select TWO.)
- AEnable MFA on the root user and store the credentials securely offline✓
- BDelete any root user access keys✓
- CUse the root user for daily administrative work
- DShare root credentials with the operations team
- EAttach an IAM policy to the root user to limit it
Correct answer: A, B — Enable MFA on the root user and store the credentials securely offline · Delete any root user access keys
Hardware or virtual MFA plus removal of root access keys are the two standard root protections. Daily use and credential sharing increase exposure, and IAM policies cannot restrict the root user within its own account.
AWS — Root user best practicesA security review finds a security group allowing 0.0.0.0/0 on port 3389. What is the most appropriate remediation?
- ARemove the rule and provide administrative access through Session Manager or a restricted source range✓
- BChange the port to a non-standard high port
- CAdd a network ACL rule allowing the same traffic
- DEnable detailed monitoring on the instance
Correct answer: A — Remove the rule and provide administrative access through Session Manager or a restricted source range
Remote administration exposed to the entire internet must be closed, and access should come through an authenticated, logged channel or a tightly scoped source range. Changing the port is security by obscurity, an ACL that allows the same traffic changes nothing, and monitoring does not remove exposure.
AWS — Security group best practicesAn S3 bucket policy allows a role from another account, but the caller still receives Access Denied. The role's IAM policy allows s3:GetObject. What else should be checked?
- AWhether the bucket's Block Public Access, an SCP, a VPC endpoint policy, or the KMS key policy is denying the request✓
- BWhether the bucket has versioning enabled
- CWhether the object is stored in S3 Standard
- DWhether the bucket has a lifecycle rule
Correct answer: A — Whether the bucket's Block Public Access, an SCP, a VPC endpoint policy, or the KMS key policy is denying the request
Cross-account S3 access passes through several policy layers, and an explicit deny anywhere, including an SCP, an endpoint policy, or the KMS key policy for encrypted objects, wins. Versioning, storage class, and lifecycle rules do not affect authorisation.
AWS — Troubleshoot access denied in Amazon S3A company wants to detect unusual API call volume that may indicate account compromise, without writing detection rules. Which feature provides this?
- ACloudTrail Insights✓
- BCloudTrail log file validation
- CAWS Config recorder
- DVPC Flow Logs
Correct answer: A — CloudTrail Insights
CloudTrail Insights baselines normal API call rates and error rates and raises events when activity deviates, with no rules to author. Log file validation proves integrity, the Config recorder tracks configuration items, and Flow Logs capture network metadata.
AWS — CloudTrail InsightsA governance requirement states that security tooling accounts must be separate from workload accounts. What is the primary security benefit?
- AA compromised workload account cannot tamper with the security tooling, logs, or findings✓
- BIt reduces the number of IAM policies required
- CIt removes the need for encryption
- DIt lowers data transfer costs
Correct answer: A — A compromised workload account cannot tamper with the security tooling, logs, or findings
Separation puts the audit trail and detection capability outside the blast radius of a compromised workload, which is what makes the evidence trustworthy. It does not reduce policy count, remove the need for encryption, or lower transfer cost.
AWS — AWS Security Reference ArchitectureA web application must be protected from SQL injection and cross-site scripting attempts at the edge. Which service should be attached to the CloudFront distribution?
- AAWS WAF with managed rule groups✓
- BAWS Shield Standard alone
- CA network ACL on the origin subnet
- DSecurity groups on the origin instances
Correct answer: A — AWS WAF with managed rule groups
WAF inspects HTTP request content and managed rule groups include signatures for injection and scripting attacks. Shield Standard mitigates network and transport layer DDoS, and network ACLs and security groups filter by address and port without understanding request payloads.
AWS — AWS WAF managed rulesWhich two configurations ensure CloudTrail captures S3 object-level read and write activity? (Select TWO.)
- AEnable data events for the S3 bucket on the trail✓
- BSelect the appropriate read and write event selectors for the resources✓
- CEnable management events only
- DEnable S3 server access logging instead and disable the trail
- EEnable CloudTrail Insights
Correct answer: A, B — Enable data events for the S3 bucket on the trail · Select the appropriate read and write event selectors for the resources
Object-level activity is a data event, and you must both enable data events for the bucket and choose read, write, or both in the event selector. Management events cover control plane calls only, server access logging is a separate mechanism that does not replace the trail, and Insights detects unusual API call volume rather than adding object logging.
AWS — Logging data eventsWhich configuration ensures data is encrypted in transit between an Application Load Balancer and its EC2 targets?
- AUse an HTTPS target group so the load balancer re-encrypts traffic to the targets✓
- BTerminate TLS at the load balancer and use HTTP to the targets
- CEnable EBS encryption on the targets
- DEnable access logging on the load balancer
Correct answer: A — Use an HTTPS target group so the load balancer re-encrypts traffic to the targets
End-to-end encryption requires the backend listener and target group to use HTTPS so the second hop is also protected. Plain HTTP to targets leaves that hop in the clear, EBS encryption protects data at rest, and access logs record requests.
AWS — ALB HTTPS listeners and target groupsA CI pipeline running in a third-party system must obtain AWS credentials without any stored secret. Which mechanism is correct?
- AAn IAM OIDC identity provider trusting the CI system, with a role assumed via web identity and a condition on the subject claim✓
- BAn IAM user with an access key stored in the CI secret store
- CA role with a trust policy allowing all principals
- DInstance profile credentials copied into the pipeline
Correct answer: A — An IAM OIDC identity provider trusting the CI system, with a role assumed via web identity and a condition on the subject claim
An OIDC trust with a condition pinning the repository and branch subject claim lets the pipeline exchange its own token for short-lived AWS credentials with no stored secret. A stored key is exactly what this removes, a wildcard trust policy would let anyone assume the role, and copying instance credentials is both fragile and unsafe.
AWS — OIDC federationEgress from a VPC must be restricted to an approved list of domain names, with traffic inspection. Which service provides stateful domain filtering at the VPC boundary?
- AAWS Network Firewall with a stateful rule group for domain filtering✓
- BSecurity groups with domain names in the rules
- CNetwork ACLs with wildcard entries
- DRoute 53 public hosted zones
Correct answer: A — AWS Network Firewall with a stateful rule group for domain filtering
Network Firewall supports stateful inspection including domain allow and deny lists at the VPC boundary. Security groups and network ACLs match on IP addresses and ports and cannot express domain names, and public hosted zones publish DNS records rather than filtering traffic.
AWS — Network Firewall stateful rulesA company wants engineers to receive different permissions in different accounts based on their directory group membership, managed centrally. Which approach fits?
- AIAM Identity Center permission sets assigned to directory groups per account✓
- BDuplicate IAM roles maintained by hand in each account
- COne shared IAM user with a very broad policy
- DAccount-level passwords rotated monthly
Correct answer: A — IAM Identity Center permission sets assigned to directory groups per account
Permission sets are defined once and assigned to groups per account, so Identity Center provisions and maintains the underlying roles. Hand-maintained roles drift, a shared broad user destroys least privilege and attribution, and password rotation does not address authorisation.
AWS — Permission setsWhich statement best describes the AWS shared responsibility model for Amazon RDS?
- AAWS manages the underlying host, operating system, and database engine patching, while the customer manages access control, encryption choices, and data✓
- BThe customer patches the database host operating system
- CAWS is responsible for the customer's IAM policies
- DThe customer is responsible for physical data centre security
Correct answer: A — AWS manages the underlying host, operating system, and database engine patching, while the customer manages access control, encryption choices, and data
For a managed database, AWS handles the infrastructure and engine maintenance while the customer remains responsible for who can access it, how data is encrypted, and the data itself. Customers never patch RDS hosts or secure data centres, and AWS does not author customer IAM policies.
AWS — Shared responsibility modelWhich IAM policy element restricts a permission so it applies only when the request carries a specific tag value?
- AA Condition block using aws:RequestTag or aws:ResourceTag✓
- BThe Sid element
- CThe Version element
- DThe Principal element in an identity-based policy
Correct answer: A — A Condition block using aws:RequestTag or aws:ResourceTag
Attribute-based access control is expressed through condition keys that compare request or resource tags. Sid is a statement label, Version selects the policy language version, and identity-based policies do not contain a Principal element.
AWS — Attribute-based access controlWhich practice ensures that security findings across an estate are consistently triaged and not lost?
- AAggregate findings in Security Hub with a delegated administrator, route them to a ticketing system, and track workflow status✓
- BReview findings individually in each account's console when time allows
- CTurn off noisy detection services
- DStore findings in a spreadsheet updated quarterly
Correct answer: A — Aggregate findings in Security Hub with a delegated administrator, route them to a ticketing system, and track workflow status
Central aggregation plus integration with the team's ticketing workflow gives one queue with tracked state, which is what prevents findings from being lost. Ad hoc per-account review does not scale, disabling detection hides risk, and a quarterly spreadsheet is not a triage process.
AWS — Security Hub findings workflowA company needs a single console aggregating findings from GuardDuty, Inspector, Macie, and partner tools across all accounts, with compliance standard scoring. Which service does this?
- AAWS Security Hub✓
- BAmazon Detective
- CAWS Config
- DAWS Systems Manager
Correct answer: A — AWS Security Hub
Security Hub ingests findings in a normalised format from AWS and partner services, aggregates across accounts, and scores against standards such as CIS and AWS Foundational Security Best Practices. Detective investigates individual findings, Config evaluates configuration compliance, and Systems Manager operates instances.
AWS — What is AWS Security HubA developer must be prevented from ever escalating their own privileges, even though they can create roles for applications. Which mechanism enforces this?
- AA permissions boundary required on any role the developer creates, enforced with an iam:PermissionsBoundary condition✓
- BA tag on the developer's IAM user
- CAn SNS notification when roles are created
- DMFA on the developer's console login
Correct answer: A — A permissions boundary required on any role the developer creates, enforced with an iam:PermissionsBoundary condition
Requiring a permissions boundary on created roles caps what those roles can ever do, which closes the privilege escalation path while still allowing delegation. A tag is metadata, notifications are detective, and MFA authenticates without limiting what the identity may grant.
AWS — Permissions boundariesA company must find and classify credit card numbers stored across thousands of S3 buckets. Which service is designed for this?
- AAmazon Macie✓
- BAmazon Athena
- CAWS Glue crawlers
- DAmazon Comprehend Medical
Correct answer: A — Amazon Macie
Macie uses managed data identifiers, including payment card patterns, to discover and classify sensitive data in S3 at scale. Athena queries data you already understand, Glue crawlers infer schema, and Comprehend Medical extracts clinical entities from text.
AWS — Amazon Macie managed data identifiersA regulated workload requires that key material never leave a FIPS 140-2 Level 3 validated hardware security module that the customer controls exclusively. Which service fits?
- AAWS CloudHSM✓
- BAWS KMS with an AWS managed key
- CAWS Secrets Manager
- DAWS Certificate Manager
Correct answer: A — AWS CloudHSM
CloudHSM gives dedicated, single-tenant HSMs where the customer controls the keys and AWS has no access to the key material. KMS is multi-tenant and managed by AWS, Secrets Manager stores secrets rather than providing an HSM, and ACM issues certificates.
AWS — What is AWS CloudHSMA new organisation needs a governed multi-account baseline with preventive and detective controls applied automatically. Which service provides this fastest?
- AAWS Control Tower✓
- BAWS CloudFormation alone
- CAWS Systems Manager
- DAmazon Detective
Correct answer: A — AWS Control Tower
Control Tower stands up a landing zone with organisational units, logging, and mandatory guardrails implemented as SCPs and Config rules. CloudFormation and Systems Manager are building blocks, and Detective investigates findings rather than establishing governance.
AWS — AWS Control Tower controlsWhich KMS feature allows a key to be used only by principals in a specific AWS account while the key itself lives in another account?
- AThe key policy, which is the primary access control for a KMS key, combined with grants✓
- BKey rotation
- CKey aliases
- DKey material origin set to EXTERNAL
Correct answer: A — The key policy, which is the primary access control for a KMS key, combined with grants
A KMS key's own resource policy governs who may use it, and grants provide fine-grained temporary delegation, which is how cross-account key usage is authorised. Rotation replaces backing key material, aliases are friendly names, and external key material concerns where the bytes come from.
AWS — Key policies in AWS KMSA company must terminate TLS at the load balancer using a certificate that is automatically renewed. Which service should issue the certificate?
- AAWS Certificate Manager with a public certificate✓
- BA self-signed certificate generated on the instance
- CAWS KMS asymmetric key
- DAWS Secrets Manager
Correct answer: A — AWS Certificate Manager with a public certificate
ACM issues public certificates free of charge and renews them automatically when they are attached to supported services such as ELB and CloudFront. Self-signed certificates fail browser validation, KMS keys are not TLS certificates, and Secrets Manager stores secrets rather than issuing certificates.
AWS — AWS Certificate ManagerWhich two capabilities help an incident responder determine what an assumed role actually did during a suspicious session? (Select TWO.)
- ACloudTrail events filtered by the role session name and access key ID✓
- BAmazon Detective's entity profile for the role✓
- CThe IAM policy simulator
- DAWS Budgets reports
- ERoute 53 query logs for the public zone
Correct answer: A, B — CloudTrail events filtered by the role session name and access key ID · Amazon Detective's entity profile for the role
CloudTrail records every API call with the session identity, and Detective visualises that activity over time for the entity. The policy simulator evaluates what a policy would allow rather than what happened, budget reports track cost, and public DNS query logs do not attribute AWS API activity.
AWS — CloudTrail userIdentity elementA secret used by a Lambda function must never appear in the function's environment variables or code. Which approach is best?
- ARetrieve the secret at runtime from Secrets Manager using the function's execution role, with caching✓
- BStore the secret in a plain environment variable
- CHardcode the secret and obfuscate it with base64
- DStore the secret in the function's deployment package
Correct answer: A — Retrieve the secret at runtime from Secrets Manager using the function's execution role, with caching
Fetching from Secrets Manager under the execution role keeps the value out of the function configuration and package, allows rotation, and produces an audit trail. Environment variables are visible to anyone who can read the function configuration, and base64 is an encoding rather than protection.
AWS — Using Secrets Manager with LambdaWhich condition key should be used to require that an API call originates from within a specific VPC endpoint?
- Aaws:SourceVpce✓
- Baws:PrincipalTag
- Caws:CurrentTime
- Daws:UserAgent
Correct answer: A — aws:SourceVpce
The aws:SourceVpce key matches the VPC endpoint identifier the request came through, which is how buckets are restricted to private network paths. Principal tags express identity attributes, current time bounds a validity window, and user agent is a client-supplied string that must not be trusted for authorisation.
AWS — Bucket policies for VPC endpointsWhich approach ensures container images are scanned for vulnerabilities before they can be deployed?
- AEnable enhanced scanning on the ECR repository and gate the pipeline on scan findings✓
- BScan images only after deployment to production
- CRely on the base image publisher's assurances
- DDisable image tag immutability so images can be replaced quickly
Correct answer: A — Enable enhanced scanning on the ECR repository and gate the pipeline on scan findings
Scanning at push time and failing the pipeline on unacceptable findings keeps vulnerable images out of production. Post-deployment scanning discovers issues too late, publisher assurances are not verification, and mutable tags make it harder to know what is actually running.
AWS — Image scanning in Amazon ECRDuring an incident, a responder must revoke all active sessions issued by a role immediately, including ones with hours of validity remaining. What is the correct approach?
- AAttach an inline deny policy to the role with a condition on aws:TokenIssueTime before the current time✓
- BDelete the role and recreate it with the same name later
- CRotate the account root password
- DChange the role's maximum session duration
Correct answer: A — Attach an inline deny policy to the role with a condition on aws:TokenIssueTime before the current time
AWS documents a revoke-sessions policy that denies all actions for credentials issued before a chosen timestamp, which invalidates existing sessions without breaking future ones. Deleting the role disrupts legitimate workloads, the root password is unrelated, and the session duration setting only affects sessions issued after the change.
AWS — Revoking IAM role temporary credentialsA KMS key policy grants a role kms:Decrypt but calls still fail with AccessDenied. The role's IAM policy has no KMS statement. What is the likely cause?
- AThe identity-based policy must also allow the KMS action unless the key policy delegates to IAM✓
- BKMS keys cannot be used by roles
- CThe key must be rotated before use
- DThe key must be in the same subnet as the caller
Correct answer: A — The identity-based policy must also allow the KMS action unless the key policy delegates to IAM
Access to a KMS key requires both the key policy and, unless the key policy delegates authority to IAM for the account, an identity-based allow on the calling principal. Roles can absolutely use KMS keys, rotation is unrelated to authorisation, and KMS keys are not subnet-scoped.
AWS — Determining access to a KMS keyWhich service finds resources shared with external principals, such as buckets or roles accessible from outside the organisation?
- AIAM Access Analyzer external access findings✓
- BAWS Trusted Advisor cost checks
- CAmazon Inspector
- DAWS Config recorder
Correct answer: A — IAM Access Analyzer external access findings
Access Analyzer uses automated reasoning over resource policies to report which resources are reachable by principals outside the defined zone of trust. Trusted Advisor cost checks look at spend, Inspector scans workloads for vulnerabilities, and the Config recorder captures configuration state without this analysis.
AWS — IAM Access Analyzer findingsWhich approach best implements least privilege for a new application role over time?
- AStart from a narrow policy, then use IAM Access Analyzer policy generation from CloudTrail activity to refine it✓
- BStart with AdministratorAccess and never revisit it
- CGrant wildcard actions on wildcard resources for convenience
- DAttach the same policy used by the operations team
Correct answer: A — Start from a narrow policy, then use IAM Access Analyzer policy generation from CloudTrail activity to refine it
Generating a policy from observed CloudTrail activity produces a grant that matches what the workload actually does, and starting narrow avoids over-permission in the meantime. Administrator access, wildcards, and reusing an unrelated team's policy all grant far more than the application needs.
AWS — Generate policies based on access activityAn organisation needs to ensure encryption at rest is enabled for every new RDS instance, blocking creation otherwise. Which control type is required?
- AA preventive control such as an SCP condition on the create API✓
- BA detective control such as a Config rule alone
- CA responsive control such as an incident runbook
- DA documentation control such as a policy wiki
Correct answer: A — A preventive control such as an SCP condition on the create API
Blocking creation is by definition preventive, and an SCP condition on the API call is the mechanism that stops the action. A Config rule reports non-compliance after the fact, runbooks respond to incidents, and documentation does not enforce anything.
AWS — Control types in AWS Control TowerWhich two are appropriate ways to demonstrate compliance evidence to an external auditor? (Select TWO.)
- AProvide AWS Artifact reports covering AWS's own controls✓
- BProvide AWS Audit Manager assessment reports for the customer-managed controls✓
- CGrant the auditor administrator access to production accounts
- DEmail screenshots of the console
- EShare the root user credentials for verification
Correct answer: A, B — Provide AWS Artifact reports covering AWS's own controls · Provide AWS Audit Manager assessment reports for the customer-managed controls
Artifact supplies AWS's third-party attestations and Audit Manager collects evidence continuously for the customer's own control set. Granting administrator access, emailing screenshots, and sharing root credentials are unacceptable in an audit context.
AWS — AWS Audit ManagerLong-lived access keys for an IAM user have been posted publicly. What is the correct immediate action?
- ADeactivate and delete the exposed keys, then review CloudTrail for their use and rotate anything they touched✓
- BAdd an IP condition to the user's policy and keep the keys
- CEnable MFA on the user and keep the keys active
- DWait for the automatic 90-day rotation
Correct answer: A — Deactivate and delete the exposed keys, then review CloudTrail for their use and rotate anything they touched
An exposed credential must be revoked immediately and its activity reviewed to scope the incident. IP conditions and MFA do not stop a key already in an attacker's hands from being used within allowed contexts, and there is no automatic rotation of IAM access keys.
AWS — What to do if you expose an access keyWhat is the effect of an explicit Deny in any applicable policy when other policies contain an Allow?
- AThe request is denied, because an explicit deny always overrides an allow✓
- BThe most specific resource ARN wins
- CThe policy attached most recently wins
- DThe request succeeds with reduced permissions
Correct answer: A — The request is denied, because an explicit deny always overrides an allow
AWS policy evaluation applies explicit deny with absolute precedence over any allow, regardless of specificity or attachment order. There is no partial success outcome in this evaluation.
AWS — Policy evaluation logicFollowing a ransomware event, a company must restore S3 objects that were overwritten with encrypted versions. Which prior configuration makes this possible?
- AS3 Versioning, so previous object versions can be restored✓
- BS3 Transfer Acceleration
- CS3 Intelligent-Tiering
- DS3 Requester Pays
Correct answer: A — S3 Versioning, so previous object versions can be restored
Versioning keeps the previous object versions when a key is overwritten, so the clean copies remain retrievable. Transfer Acceleration speeds transfers, Intelligent-Tiering optimises storage cost, and Requester Pays changes who is billed.
AWS — Using versioning in S3 bucketsWhich statement about S3 default encryption is correct?
- AAll new objects are encrypted at rest by default with server-side encryption, and a bucket can specify SSE-KMS with a chosen key✓
- BObjects are only encrypted if the client sends an encryption header
- CDefault encryption applies retroactively to existing objects
- DDefault encryption prevents any cross-account access
Correct answer: A — All new objects are encrypted at rest by default with server-side encryption, and a bucket can specify SSE-KMS with a chosen key
Amazon S3 applies server-side encryption to new objects by default and a bucket can be configured to use a specified KMS key instead of the S3 managed key. It does not require a client header, it does not re-encrypt objects that already exist, and it is unrelated to cross-account authorisation.
AWS — Default encryption for S3 bucketsGuardDuty must be enabled for all current and future accounts in the organisation with minimal ongoing effort. What should be configured?
- ADelegate a GuardDuty administrator account and enable auto-enable for new organisation accounts✓
- BEnable GuardDuty manually in each account as it is created
- CEnable GuardDuty only in the management account
- DRely on Security Hub to enable GuardDuty automatically
Correct answer: A — Delegate a GuardDuty administrator account and enable auto-enable for new organisation accounts
A delegated administrator with auto-enable turns GuardDuty on in every member account, including ones created later, from a single place. Manual enablement drifts, the management account alone leaves members unmonitored, and Security Hub aggregates findings rather than enabling the source services.
AWS — Managing GuardDuty accounts with OrganizationsAn application must encrypt individual database fields so that even a database administrator cannot read them. Which approach fits?
- AClient-side envelope encryption with the AWS Encryption SDK and a KMS key the DBA cannot use✓
- BEnabling storage encryption on the RDS instance
- CUsing SSL for database connections
- DEnabling automated backups
Correct answer: A — Client-side envelope encryption with the AWS Encryption SDK and a KMS key the DBA cannot use
Field-level client-side encryption means the database only ever holds ciphertext, so the DBA cannot read the values without KMS permission on the key. Storage encryption is transparent to anyone with query access, TLS protects only the network hop, and backups do not restrict readability.
AWS — AWS Encryption SDKWhich two statements about AWS Shield are correct? (Select TWO.)
- AShield Standard is automatically enabled for all AWS customers at no extra cost✓
- BShield Advanced adds cost protection and access to the Shield Response Team✓
- CShield Advanced replaces the need for AWS WAF entirely
- DShield Standard inspects HTTP request bodies for injection attacks
- EShield is only available in one AWS Region
Correct answer: A, B — Shield Standard is automatically enabled for all AWS customers at no extra cost · Shield Advanced adds cost protection and access to the Shield Response Team
Shield Standard is included for everyone and Shield Advanced adds DDoS cost protection plus expert response support. Shield Advanced complements rather than replaces WAF for application layer rules, Shield Standard does not perform layer 7 content inspection, and the service is not confined to a single Region.
AWS — AWS ShieldA company must apply the same WAF web ACL and security group baseline across hundreds of accounts and have new resources covered automatically. Which service does this?
- AAWS Firewall Manager✓
- BAWS WAF in each account separately
- CAmazon GuardDuty
- DAWS Config aggregator
Correct answer: A — AWS Firewall Manager
Firewall Manager applies WAF, Shield Advanced, security group, and Network Firewall policies organisation-wide and automatically covers newly created in-scope resources. Per-account WAF configuration drifts, GuardDuty detects threats, and a Config aggregator only collects data.
AWS — AWS Firewall ManagerA company must define which AWS Regions may be used and enforce it centrally. Which mechanism applies?
- AAn SCP using the aws:RequestedRegion condition key✓
- BA tag policy
- CA backup policy
- DA CloudWatch alarm on Region usage
Correct answer: A — An SCP using the aws:RequestedRegion condition key
The aws:RequestedRegion condition in an SCP denies API calls targeting unapproved Regions for all principals in scope. Tag policies standardise tags, backup policies govern AWS Backup, and an alarm only observes.
AWS — SCP examples for RegionsWhich AWS Config capability lets a security team evaluate all accounts in an organisation against a packaged set of security rules in one deployment?
- AConformance packs deployed through the organisation✓
- BA single custom rule in the management account
- CConfig aggregators only
- DConfig advanced queries only
Correct answer: A — Conformance packs deployed through the organisation
Conformance packs bundle Config rules and remediation actions into a single deployable unit that can be applied organisation-wide. A rule in one account evaluates only that account, aggregators collect data without evaluating, and advanced queries search existing configuration data.
AWS — Conformance packsAn organisation must ensure that responders in the security account can act in member accounts during an incident without permanent privileges there. Which design fits?
- AA break-glass incident response role in each member account, trusted by the security account and monitored by CloudTrail alerts on assumption✓
- BAdministrator IAM users in every member account held by the security team
- CSharing the member account root credentials in a vault
- DAn SCP that denies all actions during incidents
Correct answer: A — A break-glass incident response role in each member account, trusted by the security account and monitored by CloudTrail alerts on assumption
A trusted role assumed on demand gives privileged access only when needed and generates an auditable signal every time it is used. Standing administrator users and shared root credentials are permanent risk, and an SCP that denies everything would block the responders as well.
AWS — Cross-account roles for incident responseAfter a GuardDuty finding about credential use from an unusual location, an analyst needs to see the full behavioural history of that IAM role over the past month. Which service accelerates this investigation?
- AAmazon Detective✓
- BAWS Trusted Advisor
- CAWS Budgets
- DAmazon Macie
Correct answer: A — Amazon Detective
Detective builds a behaviour graph from CloudTrail, VPC Flow Logs, and GuardDuty findings so you can pivot through an entity's activity over time without writing queries. Trusted Advisor gives best-practice checks, Budgets tracks spend, and Macie classifies sensitive data in S3.
AWS — What is Amazon DetectiveWhich two are appropriate uses of an IAM role rather than an IAM user? (Select TWO.)
- AAn EC2 instance that needs to call AWS APIs✓
- BA federated employee signing in through the corporate identity provider✓
- CA long-lived credential embedded in a public GitHub repository
- DA password shared between contractors
- EStoring an access key in an unencrypted config file
Correct answer: A, B — An EC2 instance that needs to call AWS APIs · A federated employee signing in through the corporate identity provider
Roles deliver temporary credentials to workloads and federated humans, which removes long-lived secrets from both paths. The remaining options describe credential handling practices that should never occur.
AWS — IAM rolesAn IAM policy allows an action but a service control policy on the account does not include it. What is the result?
- AThe action is denied because the SCP does not permit it✓
- BThe action is allowed because the IAM policy grants it
- CThe action is allowed only from the console
- DThe action requires MFA to proceed
Correct answer: A — The action is denied because the SCP does not permit it
An action must be permitted by both the SCP ceiling and an IAM policy, so anything outside the SCP is denied regardless of IAM grants. Console versus API access does not change policy evaluation, and MFA is a separate condition.
AWS — Policy evaluation logicA responder needs a forensic copy of an encrypted EBS volume in a separate forensics account. What must be in place?
- AThe snapshot must be shared and re-encrypted with a KMS key the forensics account can use✓
- BThe volume must be made unencrypted first
- CThe forensics account must be the organisation management account
- DThe snapshot must be made public
Correct answer: A — The snapshot must be shared and re-encrypted with a KMS key the forensics account can use
Snapshots encrypted with a customer managed key can be shared and copied, but the destination account needs permission on a key it can use, so the copy is re-encrypted appropriately. Decrypting the volume weakens protection, the forensics account need not be the management account, and public snapshots would expose the evidence.
AWS — Share an encrypted snapshotWhich GuardDuty finding type would indicate that an EC2 instance is communicating with a known command and control server?
- AA Backdoor or CryptoCurrency finding based on DNS and network activity✓
- BAn S3 public access finding
- CAn IAM policy compliance finding
- DA cost anomaly finding
Correct answer: A — A Backdoor or CryptoCurrency finding based on DNS and network activity
GuardDuty analyses DNS queries and VPC Flow Logs against threat intelligence, so contact with known malicious infrastructure surfaces as a backdoor or cryptocurrency finding type. Public access and policy compliance are Config or Access Analyzer concerns, and cost anomalies come from the billing service.
AWS — GuardDuty finding types for EC2A security engineer must alert on a specific pattern appearing in application logs already streaming to CloudWatch Logs. Which feature turns that pattern into an alarm?
- AA metric filter that publishes a custom metric, with a CloudWatch alarm on it✓
- BA subscription filter to an S3 bucket
- CA CloudWatch dashboard widget
- DA log group retention policy
Correct answer: A — A metric filter that publishes a custom metric, with a CloudWatch alarm on it
Metric filters count matching log events into a CloudWatch metric that an alarm can then evaluate and notify on. Subscription filters stream data elsewhere without alarming, dashboards display, and retention policies control how long logs are kept.
AWS — Creating metric filtersA workload requires that all traffic between two subnets be inspected by a third-party virtual appliance. Which service inserts the appliance transparently?
- AGateway Load Balancer with GENEVE encapsulation and endpoint routing✓
- BApplication Load Balancer with a target group
- CNetwork Load Balancer with cross-zone balancing
- DCloudFront with a custom origin
Correct answer: A — Gateway Load Balancer with GENEVE encapsulation and endpoint routing
Gateway Load Balancer plus its VPC endpoints let route tables steer traffic through a scalable fleet of inline appliances without changing the application. The application and network load balancers terminate or proxy connections to targets, and CloudFront is a content delivery service.
AWS — Gateway Load BalancerA company wants automated containment when GuardDuty reports a compromised instance. Which architecture achieves this?
- AAn EventBridge rule on the GuardDuty finding invoking a Systems Manager Automation runbook or Lambda that applies an isolation security group✓
- BAn SNS email to the on-call engineer only
- CA daily script that reads Security Hub findings
- DA CloudWatch dashboard with a red widget
Correct answer: A — An EventBridge rule on the GuardDuty finding invoking a Systems Manager Automation runbook or Lambda that applies an isolation security group
GuardDuty findings are delivered to EventBridge, which can trigger automation that quarantines the instance in seconds. Email alone waits on a human, a daily script is far too slow, and a dashboard does not act.
AWS — Automating GuardDuty responsesA team must detect software vulnerabilities and unintended network exposure in EC2 instances and container images continuously. Which service fits?
- AAmazon Inspector✓
- BAmazon GuardDuty
- CAWS Firewall Manager
- DAWS Artifact
Correct answer: A — Amazon Inspector
Inspector continuously scans EC2 instances, ECR images, and Lambda functions for CVEs and evaluates network reachability. GuardDuty detects active threats rather than vulnerabilities, Firewall Manager centralises firewall policy, and Artifact provides compliance documents.
AWS — What is Amazon InspectorAn engineer needs to know whether a proposed IAM policy change would still allow a specific action on a specific resource. Which tool answers this before deployment?
- AThe IAM policy simulator✓
- BCloudTrail Insights
- CAWS Config timeline
- DCost Explorer forecasting
Correct answer: A — The IAM policy simulator
The policy simulator evaluates a policy against specified actions and resources and reports allow or deny with the deciding statement, which is a pre-deployment check. CloudTrail Insights and the Config timeline describe what already happened, and Cost Explorer forecasts spend.
AWS — Testing IAM policies with the policy simulatorInstances in a private subnet must reach AWS APIs without a NAT gateway, and API calls must be restricted to a specific organisation. Which combination applies?
- AInterface VPC endpoints with an endpoint policy using the aws:PrincipalOrgID condition✓
- BA NAT gateway with a restrictive security group
- CAn internet gateway plus a network ACL
- DA Transit Gateway with default routes
Correct answer: A — Interface VPC endpoints with an endpoint policy using the aws:PrincipalOrgID condition
Interface endpoints keep API traffic private, and an endpoint policy can restrict which principals and organisations may use the endpoint. NAT gateways and internet gateways route to the internet, and a Transit Gateway routes between networks rather than restricting AWS API access.
AWS — VPC endpoint policiesA bastion-free administrative access model is required for EC2 instances in private subnets, with full session logging. Which option meets this?
- AAWS Systems Manager Session Manager with session logging to S3 or CloudWatch Logs✓
- BAn SSH bastion host with a public IP and key pairs
- COpening port 22 to the corporate CIDR
- DA Site-to-Site VPN plus shared SSH keys
Correct answer: A — AWS Systems Manager Session Manager with session logging to S3 or CloudWatch Logs
Session Manager connects through the SSM agent with no inbound ports, no bastion, and IAM-controlled access, and it can record full session transcripts. Every other option keeps SSH, key distribution, and an exposed listener in the design.
AWS — Session Manager loggingAn application needs to be protected against a volumetric DDoS attack with cost protection for the resulting scaling charges. Which option applies?
- AAWS Shield Advanced✓
- BAWS WAF rate-based rules only
- CAmazon GuardDuty
- DAWS Budgets alerts
Correct answer: A — AWS Shield Advanced
Shield Advanced provides enhanced DDoS protection plus cost protection credits for scaling charges caused by an attack, along with response team access. Rate-based WAF rules help at layer 7 but carry no cost protection, GuardDuty detects rather than mitigates, and Budgets only alert on spend.
AWS — Shield Advanced benefitsAn EC2 instance is suspected of being compromised. Which sequence best preserves evidence while containing the threat?
- AIsolate with a restrictive security group, take an EBS snapshot and memory capture, then investigate before terminating✓
- BTerminate the instance immediately to stop the attack
- CReboot the instance and monitor
- DDetach the IAM role and leave the instance in the production security group
Correct answer: A — Isolate with a restrictive security group, take an EBS snapshot and memory capture, then investigate before terminating
Containment before eradication preserves the forensic record: an isolation security group cuts communication while snapshots and memory capture retain evidence. Terminating destroys evidence, rebooting may wipe volatile memory and does not contain anything, and leaving the instance reachable fails to contain it.
AWS — Incident response for EC2Which two are true about VPC security groups? (Select TWO.)
- AThey are stateful, so return traffic for an allowed connection is permitted automatically✓
- BThey can reference another security group as a source✓
- CThey support explicit deny rules
- DThey are evaluated in rule number order
- EThey apply at the subnet level
Correct answer: A, B — They are stateful, so return traffic for an allowed connection is permitted automatically · They can reference another security group as a source
Security groups track connection state and can use another security group as a source, which is how tiered architectures are expressed without hardcoding addresses. They contain allow rules only, have no rule ordering, and attach to network interfaces rather than subnets.
AWS — Security groupsWhich practice most improves the speed of incident response before an incident occurs?
- APre-provisioned forensic tooling, a dedicated response role, and rehearsed runbooks✓
- BStoring the response plan only in a shared document
- CGranting responders administrator access permanently
- DDisabling logging in production to reduce noise
Correct answer: A — Pre-provisioned forensic tooling, a dedicated response role, and rehearsed runbooks
Response speed comes from having access, tools, and practised procedures ready before the event. A document nobody has rehearsed slows the response, standing administrator access is a risk in itself, and disabling logging destroys the evidence responders depend on.
AWS — Security incident response guideA security team must be alerted within minutes when an IAM user creates an access key for the root user in any account. Which combination provides the fastest detection?
- AAn EventBridge rule matching the CloudTrail event, targeting an SNS topic✓
- BA weekly review of CloudTrail log files in S3
- CAn AWS Config rule evaluated once every 24 hours
- DA Cost Explorer anomaly alert
Correct answer: A — An EventBridge rule matching the CloudTrail event, targeting an SNS topic
CloudTrail management events flow to EventBridge in near real time, so a rule matching the API call can notify within minutes. Weekly log review and daily Config evaluation are far too slow, and cost anomaly detection watches spend rather than identity actions.
AWS — CloudTrail events in EventBridgeA company must guarantee that objects uploaded to a bucket are always encrypted with a specific customer managed KMS key. Which control enforces this?
- AA bucket policy denying PutObject when s3:x-amz-server-side-encryption-aws-kms-key-id does not match the approved key ARN✓
- BEnabling bucket versioning
- CEnabling S3 Transfer Acceleration
- DAdding a lifecycle rule
Correct answer: A — A bucket policy denying PutObject when s3:x-amz-server-side-encryption-aws-kms-key-id does not match the approved key ARN
A deny condition on the encryption header rejects any upload that does not name the approved key, which enforces the requirement at write time. Versioning, acceleration, and lifecycle rules do not constrain how objects are encrypted.
AWS — Protecting data with server-side encryptionA backup vault must prevent anyone, including administrators, from deleting recovery points before the retention period expires. Which feature provides this?
- AAWS Backup Vault Lock in compliance mode✓
- BA backup plan with a long retention setting
- CAn IAM policy denying delete for one user
- DCross-Region copy of backups
Correct answer: A — AWS Backup Vault Lock in compliance mode
Vault Lock in compliance mode makes the retention policy immutable so recovery points cannot be deleted early by any principal. A retention setting alone can be changed, a single IAM deny leaves other identities unconstrained, and cross-Region copies improve durability without preventing deletion.
AWS — AWS Backup Vault LockWhich two settings help prevent accidental public exposure of S3 data across an entire organisation? (Select TWO.)
- AS3 Block Public Access enabled at the account level✓
- BAn SCP denying s3:PutBucketPublicAccessBlock changes that weaken the setting✓
- CEnabling S3 Transfer Acceleration
- DTurning on Requester Pays
- EUsing S3 Standard-IA for all objects
Correct answer: A, B — S3 Block Public Access enabled at the account level · An SCP denying s3:PutBucketPublicAccessBlock changes that weaken the setting
Account-level Block Public Access overrides permissive bucket settings, and an SCP prevents anyone from turning that protection off. Acceleration, Requester Pays, and storage class choices have no effect on public exposure.
AWS — Blocking public access to S3 storageA mobile application needs temporary AWS credentials scoped per end user after they sign in with a social identity provider. Which service fits?
- AAmazon Cognito identity pools issuing role-based temporary credentials✓
- BAn IAM user per mobile device
- CA shared access key embedded in the app binary
- DAn EC2 instance profile
Correct answer: A — Amazon Cognito identity pools issuing role-based temporary credentials
Cognito identity pools exchange a federated identity token for scoped temporary AWS credentials, which is the supported pattern for mobile clients. Per-device IAM users do not scale, embedding keys in an app leaks them immediately, and instance profiles apply to EC2 rather than mobile clients.
AWS — Amazon Cognito identity poolsWhich statement about AWS Config and CloudTrail is accurate?
- ACloudTrail records who made an API call, while Config records the resulting resource configuration state over time✓
- BConfig records API callers and CloudTrail records configuration snapshots
- CBoth services record identical data and only one is needed
- DNeither service supports multi-account aggregation
Correct answer: A — CloudTrail records who made an API call, while Config records the resulting resource configuration state over time
The two are complementary: CloudTrail answers who did what and when, and Config answers what the resource looked like at any point and whether it complies. Their roles are not reversed, they are not redundant, and both support organisation-wide aggregation.
AWS — AWS Config conceptsWhich two statements about AWS KMS key rotation are correct? (Select TWO.)
- AAutomatic rotation creates new backing key material while the key ID and ARN stay the same✓
- BData encrypted before rotation remains decryptable using the retained older key material✓
- CRotation re-encrypts all existing ciphertext automatically
- DRotation changes the key ARN, so applications must be updated
- ERotation is available for AWS owned keys under customer control
Correct answer: A, B — Automatic rotation creates new backing key material while the key ID and ARN stay the same · Data encrypted before rotation remains decryptable using the retained older key material
Rotation swaps the backing material transparently and KMS retains previous material so old ciphertext still decrypts. It does not re-encrypt existing data, the key identifiers are unchanged, and AWS owned keys are not customer-configurable.
AWS — Rotating AWS KMS keysReady to try it under exam conditions?
Reading answers is not the same as recalling them with a clock running. Take the same 65 questions as a timed mock exam — 170 minutes, no feedback until you submit, then a score broken down by exam domain so you know what to study.
Start the timed SCS-C03 test →