Pass Your Microsoft AB-900 Exam Easy!

Legal hold supersedes ordinary user deletion within its scope. That addresses mandatory legal evidence preservation, whereas the current case requires auditable record label changes.

Disposition review governs human approval before permanent deletion. That addresses reviewed record disposal, whereas the current case requires auditable record label changes.

Retention preserves records for the prescribed period before deletion. It is used for deletion following mandatory preservation; this scenario calls for auditable record label changes.

Rights management can constrain permitted operations independent of file retrieval. Its purpose is limited document use after access; the specified need is independent retention and confidentiality classification.

DLP evaluates the message content and recipient context. It would support policy-blocked sensitive outbound mail, not the stated independent retention and confidentiality classification.

File-level encryption can enforce authorized use beyond the SharePoint site boundary. It would support persistent protection on a shared document, not the stated independent retention and confidentiality classification.

Compliance Manager ties controls and improvement actions to compliance frameworks. This helps with framework-based compliance evidence instead of the required AI interaction and oversharing risk visibility.

Security score recommendations emphasize security posture rather than audit controls. The action is relevant to security control implementation priorities rather than AI interaction and oversharing risk visibility.

Classification exploration identifies where matching sensitive data resides. This helps with location of classified content instead of the required AI interaction and oversharing risk visibility.

Copilot can surface already-accessible overshared content. It is used for eliminating an inherited oversharing root cause; this scenario calls for denying retrieval of an inaccessible source.

Security trimming and individual context yield different retrieved evidence. That addresses explaining user-specific Copilot answers, whereas the current case requires denying retrieval of an inaccessible source.

Graph uses accessible work context when building grounded responses. That addresses summarizing accessible mailbox content, whereas the current case requires denying retrieval of an inaccessible source.

The relevant access risk comes from content sharing and identity authorization. It would support preventing unapproved contractor discovery, not the stated ordinary user-scoped work grounding.

Correctly mapped access lists are needed for secure external content retrieval. Its purpose is connector-grounded security trimming; the specified need is ordinary user-scoped work grounding.

Grounding checks help distinguish actual retrieved facts from generated claims. It would support verification against authorized evidence, not the stated ordinary user-scoped work grounding.

An informational label does not imply universal exclusion from Copilot. This helps with authorized summary of a labeled source instead of the required enforcement of protected document rights.

Supported Copilot data controls can restrict processing of classified items. The action is relevant to source-specific AI processing restriction rather than enforcement of protected document rights.

Labels provide consistent handling guidance for sensitive work content. This helps with end-user classification of source data instead of the required enforcement of protected document rights.

Inclusive design helps avoid excluding people because of disabilities or circumstances. That addresses usable AI experience across abilities, whereas the current case requires nondiscriminatory model behavior.

Transparency helps users understand AI use, evidence and limitations. That addresses intelligible AI decision context, whereas the current case requires nondiscriminatory model behavior.

Robustness evaluation reduces harmful surprises and inconsistent operation. It is used for dependable operation under atypical inputs; this scenario calls for nondiscriminatory model behavior.

Privacy principles reduce unnecessary processing and disclosure. Its purpose is protection of sensitive personal information; the specified need is human oversight of consequential action.

Accountability identifies responsible humans for monitoring and correction. It would support clear ownership of AI failures, not the stated human oversight of consequential action.

Evidence checking supports responsible use of generated assertions. It would support verified high-stakes summary accuracy, not the stated human oversight of consequential action.

Boundaries between content and instructions mitigate prompt-injection effects. This helps with safe handling of adversarial document text instead of the required resistance to malicious source instructions.

Prompt wording alone cannot secure a powerful tool connection. The action is relevant to enforceable permission checks for agent actions rather than resistance to malicious source instructions.

Monitoring enables investigation of attempts to misuse the assistant. This helps with evidence of AI abuse attempts instead of the required resistance to malicious source instructions.

Retention policies govern covered interaction records for required periods. That addresses mandated AI conversation retention, whereas the current case requires ongoing legal preservation of source evidence.

Content lifecycle operates on source records independent of Copilot summaries. It is used for policy-based deletion of expired source content; this scenario calls for ongoing legal preservation of source evidence.

Case-based content search preserves and locates relevant AI records. That addresses discovery of AI interaction evidence, whereas the current case requires ongoing legal preservation of source evidence.

Grounding uncertainty prevents fabricated specific claims. It would support safe answer when evidence is absent, not the stated current operational data rather than stale export.

Source stewardship reduces grounding in obsolete versions. It would support a current authoritative document reference, not the stated current operational data rather than stale export.

Citations allow checking statements against evidence. Its purpose is traceable grounded policy statement; the specified need is current operational data rather than stale export.