The documentation no longer names specific third-party detection engines. It now states only that data is available when using an approved Microsoft or third-party engine.
AI agent alerts now document unredacted prompt snippets containing sensitive data and secrets
The period’s most consequential update is Endpoint guidance for AI agent runtime protection: two pages now state that alert prompt snippets are unredacted and may expose conversations, sensitive data, and secrets. Defender XDR updates also revised the Security Alert Triage Agent’s cloud-alert coverage label and removed named third-party detection engines from synthetic-media event guidance.
- Endpoint runtime-protection guidance warns that alert prompt snippets can be unredacted
Endpoint · Endpoint protection
The configuration page now states that sensitive data and secrets may appear unredacted in prompt snippets shown in alerts. It links to current Prompt Evidence Collection details and says Intune template help text is being finalized during preview.
- Security Alert Triage Agent coverage now includes containers under a preview label
Defender XDR · Incidents and response
The documented coverage is now “High-severity cloud alerts, including containers (Preview),” alongside updated page metadata. Administrators reviewing the agent’s documented scope should use this revised wording.
- Synthetic-media event guidance now uses an approved-engine qualification
Defender XDR · Hunting and detection
The hunting documentation no longer names specific third-party detection engines. It now says data is available when using an approved Microsoft or third-party engine; the supplied guidance identifies no administrator action.
This period briefing was generated by AI from the tracked Microsoft Learn and Message Center changes.
4 updates
Microsoft Defender XDR
2 updatesSecurity Alert Triage Agent
Doc updateThe documentation now describes the coverage as “High-severity cloud alerts, including containers (Preview)” and updates the page metadata.
Microsoft Defender for Endpoint
2 updatesConfigure Ai Agent Runtime Protection
Doc updateThe page now states that sensitive data and secrets may appear unredacted in prompt snippets shown in alerts. It also links to current Prompt Evidence Collection details and notes that Intune template help text is being finalized during preview.
Ai Agent Runtime Protection Overview
Feature updateThe documentation now states that prompt snippets in alerts are not redacted and may contain conversations, sensitive data, and secrets.
