A Conditional Access policy for administrators requires the built-in Phishing-resistant MFA authentication strength. Priya, an administrator, signs in with Microsoft Authenticator phone sign-in and is blocked. Her sign-in log shows the authentication strength wasn't satisfied. Which change lets Priya meet the policy without weakening it?
ASwitch the policy to the built-in Passwordless MFA strength, which already includes Authenticator phone sign-in.
BAdd Require multifactor authentication to the same policy, next to the authentication strength control.
CHave Priya register a FIDO2 security key in Security info and use it, rather than phone sign-in, for her administrator sign-ins.
DTurn on number matching and location context for Microsoft Authenticator in the Authentication methods policy.
CCorrect. The built-in Phishing-resistant MFA strength accepts a FIDO2 security key, Windows Hello for Business or platform credential, and multifactor certificate-based authentication. Authenticator phone sign-in isn't in that list.
AIncorrect. Passwordless MFA does include Authenticator phone sign-in, but it isn't phishing-resistant. The change would let Priya in by weakening the requirement, which the scenario rules out.
BIncorrect. You can't use Require multifactor authentication and Require authentication strength together in the same policy, and plain MFA wouldn't make phone sign-in phishing-resistant anyway.
DIncorrect. Authenticator settings don't change which combinations a built-in strength allows. In the table for the Phishing-resistant MFA strength, Authenticator phone sign-in has no entry, whatever its prompt settings.
Tests
Implement Conditional Access policies: authentication strengths and which methods are phishing-resistant
A storage account's blob private endpoint lives in vnet-hub, and the private DNS zone privatelink.blob.core.windows.net is linked to vnet-hub only. VMs in the peered spoke vnet-spoke2 use Azure-provided DNS, and they resolve the storage account's name to its public IP address. What should you do?
ACreate another privatelink.blob.core.windows.net zone and link it to vnet-spoke2.
BLink the existing privatelink.blob.core.windows.net zone to vnet-spoke2.
CSet the DNS server of vnet-spoke2 to the Azure DNS address 168.63.129.16.
DEnable Use remote gateways on vnet-spoke2's peering to vnet-hub.
BCorrect. For peered virtual networks, Microsoft's model is a single private DNS zone with virtual network links to the hub and every spoke that contains clients. Once vnet-spoke2 is linked, Azure DNS answers its queries from the zone.
AIncorrect. Microsoft requires a single private DNS zone for this configuration. Separate zones with the same name for different virtual networks would need manual operations to merge their records.
CIncorrect. The spoke already uses Azure-provided DNS. What's missing is the link: DNS queries for private endpoints must originate from a virtual network linked to the private DNS zone.
DIncorrect. The page's fix for peered virtual networks is to add virtual network links to the private DNS zone. Peering settings don't change which private DNS zones a virtual network resolves.
Tests
Private endpoints: DNS resolution from peered spoke virtual networks
Not a preview. This is question 3.15 with its full explanation, so you can judge the depth before paying.
Q 3.15 · Secure computedifficulty 2 / 5
A security engineer must turn on threat protection for the Azure OpenAI deployments in one Azure subscription, so that jailbreak attempts and credential leakage in model responses raise security alerts. Where does the engineer turn it on?
ADefender portal, under Security for AI real-time protection
BFoundry portal, on the AI Gateway pane for the resource
CPurview portal, on the DSPM setup tasks page
DDefender for Cloud environment settings, AI services plan
DCorrect. In the Azure portal, open Microsoft Defender for Cloud > Environment settings, select the subscription, and on the Defender plans page toggle AI services to On.
AIncorrect. Threat protection for AI services is a Defender for Cloud plan, turned on per subscription in the Azure portal under Environment settings > Defender plans. Real-time protection settings in the Defender portal aren't where that plan is enabled.
BIncorrect. The plan isn't a setting on the Foundry resource. It's turned on per subscription in Defender for Cloud's Environment settings, on the Defender plans page.
CIncorrect. The plan isn't enabled from Microsoft Purview. It's turned on per subscription in Defender for Cloud's Environment settings, on the Defender plans page.
Tests
Enable Defender for AI Services in Defender for Cloud: where the plan is enabled
Not a preview. This is question 3.18 with its full explanation, so you can judge the depth before paying.
Q 3.18 · Secure computedifficulty 3 / 5
A Foundry agent answers questions from supplier PDFs it retrieves at run time. In a red-team test, hidden instructions planted in one PDF make the agent ignore its system instructions. The agent's guardrail uses the default Prompt Shields settings. Which change addresses this attack?
ATurn on document attack detection in the guardrail
BRaise the Violence blocking level to High for the agent
CAdd a groundedness control to the guardrail
DTurn on user prompt attack detection in the guardrail
ACorrect. Prompt Shields detect two attack types. Attacks hidden in document content, such as emails, web pages or files, are document attacks, and their detection is off by default and must be enabled in the guardrail's controls.
BIncorrect. Hidden instructions in a document are a document attack, which Prompt Shields detect. A harmful-content level judges whether content is harmful, and Prompt Shields' job, detecting manipulation, is a different check.
CIncorrect. Detecting instructions planted in documents is the job of Prompt Shields' document attack detection. Microsoft is explicit that Prompt Shields don't judge groundedness, and a grounding check is a different control that doesn't look for manipulation attempts.
DIncorrect. User prompt attack detection is already on by default, and it analyzes what the user types. The attack here arrived inside a retrieved document, not in the user's prompt.
Tests
Configure guardrails for agent security in Foundry: enabling Prompt Shields for document attacks
Tier 1 analysts at Contoso must be able to assign Sentinel incidents to themselves, change incident status and add comments. They must not create analytics rules, install content hub solutions or run playbooks. Which built-in role should they get on the Sentinel resource group?
AMicrosoft Sentinel Automation Contributor
BMicrosoft Sentinel Responder
CMicrosoft Sentinel Contributor
DMicrosoft Sentinel Reader
BCorrect. Responder has all Reader permissions plus managing incidents, and nothing more: no content hub, no analytics rules and no running playbooks.
AIncorrect. This role lets Microsoft Sentinel add playbooks to automation rules and isn't used for user accounts.
CIncorrect. Contributor adds installing and updating solutions and creating and editing resources such as analytics rules, which the analysts must not have.
DIncorrect. Reader can view data, incidents, workbooks and recommendations but can't manage incidents, so the analysts couldn't assign or update them.
Tests
Assign roles in Microsoft Sentinel: least privilege for incident handling