A risky AI use alert gets an owner
The sample response names who reviews and who may act.
Coming
See every demonstrationFREE CHECK
AI security and agent governance check
Inspect data reach, action gates, monitoring, and recovery.
01 Source02 Boundary03 Human decision04 Evidence
WHAT TO REVIEW
Four questions for the team that owns the decision.
- Do agents retrieve only what the caller can access?
- Are connected tools and their permissions inventoried?
- Does every production action have a human approval point?
- Can you revoke an agent and verify that access ended?
Take the PDF with you.
Enter a work email, company and role to request the PDF. This action records an SIT inquiry for follow-up. Marketing emails require a separate choice.
Download the PDF directlyTHE CHECK
What does the AI security and agent governance check test?
Whether agents retrieve only what the caller can access, whether connected tools and their permissions are inventoried, whether every production action has a human approval point, and whether you can revoke an agent and verify that access ended.
Four questions on data reach, action gates, monitoring and recovery, with a PDF you can download directly and no email required.
THE MICROSOFT INSIGHT
Can an agent's action be checked before it runs?
Yes: Microsoft Defender real-time protection evaluates Agent 365 tool invocations made through Work IQ MCP before they run, and custom rules can block matching actions.
Microsoft Defender real-time protection evaluates Agent 365 tool invocations made through Work IQ MCP before they run; the default rule audits and custom rules can block matching actions. (Microsoft Learn, checked Oct 31, 2026)
A blocking rule turns a bad tool call into an alert. Without it, the tool call runs and the alert is the outcome.
THE METHOD
How does a four-question check lead to a boundary?
Each question is one edge of the boundary: what the agent can reach, what it can use, when a person steps in, and how it is switched off.
- Reach: retrieval limited to what the caller may already open.
- Tools: connected tools and their permissions inventoried, with owners.
- Gate: every production action has a human approval point.
- Off switch: revoke and verify, so an agent can be stopped and the stop proven.
THE NEXT RUNG
What is the AI Data Readiness and Agent Governance Assessment?
A read-only engagement that tests the four edges inside your tenant and produces the inventory, the first permission rows and the Tested or Assumed record.
Microsoft co-markets a related engagement as the Security for AI Assessment. The scope here is set with you and quoted after a scoping call. Microsoft-funded where you qualify.
Test the edges before the rollout and the record is a design. Test them after and the record is an incident timeline.
THE SIGNAL
How do you see risky AI use before it becomes an incident?
Microsoft Purview Insider Risk Management ships a Risky AI usage policy template that scores prompts and responses containing sensitive information and browsing to generative AI sites.
Microsoft Purview Insider Risk Management ships a Risky AI usage policy template that scores prompts and responses containing sensitive information in Microsoft 365 Copilot and agents, and browsing to generative AI sites. (Microsoft Learn, checked Oct 31, 2026)
With the policy, risky use has an owner. Without it, risky use has a headline.
AFTER TWO WEEKS
What will be working after two weeks?
- You hold the PDF and the four edges are marked tested or assumed.
- The connected tools are listed with owners, or the missing inventory is named as the gap.
- The first production action without an approval point is identified.
- You have booked, or decided not to book, the read-only assessment.
HONEST ABOUT LIMITS
What this does not do yet
It does not detect every attack on an agent with a control that is generally available today. Microsoft Defender threat detection for Agent 365 managed agents, in public preview, raises near real time alerts for jailbreak attempts, indirect prompt injection, credential leakage and suspicious access. (Microsoft Learn, checked Nov 4, 2026, preview) The preview is labeled, and the gate and the blocking rule carry the design meanwhile.
It does not test the edges: the check finds the assumption, the assessment tests it.
It does not create a lead unless you ask for the PDF by email and say so.
THE ONE THING TO REMEMBER
Reach, tools, gate, off switch: four edges of a boundary you can draw before the audit.
Said another way: an agent you cannot switch off is not governed.
Said a third way: find the assumed edge before the examiner finds it for you.
SHOW YOUR WORK
See the control path on a fictional bank.
DIRECT ANSWERS
Questions before you decide.
Does the check cover vendor agents?
Yes. Any agent that can reach your data or act in your tenant is in scope, including vendor agents, which join Microsoft Agent 365 through registry sync or the SDK.
Is the PDF gated?
No. Download it directly. Asking for it by email is optional and creates an inquiry; marketing email is a separate box.
What is the difference between this and the Agent 365 readiness check?
This check draws the security boundary: reach, tools, gate, off switch. The Agent 365 check tests governance: owner, identity, gate, trace. Many teams take both.
What comes after the check?
The AI Data Readiness and Agent Governance Assessment, read-only, inside your tenant, with the inventory and the first permission rows as the output.
How is the assessment priced?
Quoted after a scoping call, scoped separately from remediation. Microsoft-funded where you qualify.
NEXT STEP
Start with a boundary you can explain.
Choose a focused check, or speak with the team about the workflow you need to improve.