From the course: Microsoft AI Business Professional (AB-730) Cert Prep by Microsoft Press
Unlock this course with a free trial
Join today to access over 26,500 courses taught by industry experts.
Recognize and mitigate risks to sensitive data
From the course: Microsoft AI Business Professional (AB-730) Cert Prep by Microsoft Press
Recognize and mitigate risks to sensitive data
It's important to consider how we handle sensitive data when using Copilot. One immediate concern is what users type into Copilot. The prompt itself can contain sensitive information. For example, an employee might copy-paste a spreadsheet of salaries or a list of customer contact details into Copilot to get an e-mail summary or an e-mail drafted. Once you do that, all those details become part of Copilot's processing, and the AI is going to use them to generate a response. And it might even include some of those specifics in its output. So for instance, mentioning a particular salary or customer's name if you asked it. And so that means there's a risk of leaking sensitive data via the AI, because if then the user shares or sends the AI's output without realizing it contains confidential tidbits from the prompt, it could accidentally expose information. And moreover, prompts and outputs might be stored for auditing, meaning there's a record of that sensitive data moving through the…
Contents
-
-
-
-
(Locked)
Learning objectives49s
-
(Locked)
Identify common risks, including fabrications, prompt injection, and over-reliance8m 25s
-
(Locked)
Select verification steps appropriate to the task, including citation checks and human review11m 21s
-
(Locked)
Recognize and mitigate risks to sensitive data11m 19s
-
(Locked)
Understand how data protection restricts prompt results9m 31s
-
(Locked)
-
-
-
-
-
-