About this guide
Microsoft Copilot Chat is a web-based AI chat assistant available to the WSU community through Microsoft 365. This guide provides practical, role-based prompts and safety tips to help students, faculty, staff, and researchers use Copilot Chat effectively and responsibly.
The guide is organized into three sections: a brief setup overview, core safety principles for all users, and a prompt library grouped by audience.
Shaded boxes in the guide contain prompts you can copy directly into Copilot Chat; simply replace the [bracketed placeholders] with your own details. Each prompt also includes a real-world example to show how it can be adapted.
Download a PDF version of this guide.
Table of Contents
Getting started
Sign in first
Open Copilot Chat and sign in with your WSU account. You can access Copilot Chat through a web browser, the Microsoft 365 Copilot app, or the Copilot icon in the Microsoft Edge sidebar.
Look for the green shield
When you are signed in with your WSU account, Copilot Chat includes enterprise data protection. This is indicated by a green shield icon near the top of the chat window. Enterprise data protection means your prompts and responses stay within the Microsoft 365 environment and are not used to train the underlying AI models.
Who can see your prompts
Access to Copilot Chat data is limited to a small number of authorized WSU IT administrators. Access is granted only for legitimate institutional or legal purposes, never to monitor everyday use.
What it means for your data
Because you are working in a protected environment, you may use institutional information you are authorized to access as long as you have permission from the data owner and follow all applicable policies (FERPA, contract requirements, etc.).
Enterprise data protection applies to how your prompts are handled; it does not grant permission to use data you are not otherwise allowed to access or handle.
Quick reminders
Before sharing information with Copilot Chat:
- Confirm the green shield icon is visible and that you are signed in with your WSU account. Personal Microsoft accounts do not include these protections and should not be used for WSU work.
- Make sure you are authorized to use information with Copilot Chat and that the data owner has approved its use. When possible, use the de-identifying prompts in this guide instead of real data.
One more important note: Copilot Chat uses information from the open web. It does not access your WSU files, email, or Teams chats. To work directly with your documents and email, a separate paid Microsoft 365 Copilot license is required.
Top 10 high-risk misuse patterns
Even in a protected environment, certain practices can create risks. Watch for the following:
- Entering data you are not authorized to access or use with AI.
- Using Copilot while signed out or in a personal Microsoft account.
- Treating AI responses as accurate without human verification.
- Using AI to make decisions about people (grading, hiring, discipline, financial aid, admissions, accommodations, etc.).
- Trusting unverified citations and research results.
- Using AI to bypass learning objectives or academic integrity rules.
- Relying on a single response instead of refining and iterating.
- Allowing AI to introduce bias, exclusion, or inequity into communications or processes.
- Sharing AI-generated policy or compliance interpretations without verifying official sources.
- Failing to disclose meaningful AI assistance when expected in academic, research, or administrative work.
How to use the prompt library
- Copy the prompt text from the shaded box into Copilot Chat.
- Replace each [bracketed placeholder] with your specific details.
- Adapt the wording using the example beneath the prompt.
- Add one or more universal safety add-ons (later in this guide) to the bottom of any prompt to improve privacy and accuracy.
A helpful mindset: treat Copilot as a fast, capable assistant and treat yourself as the editor and decision-maker. The following prompts are written to support that balance.
Core safety principles for everyone
Three key practices apply to all users and tasks. Each can be used directly as a prompt.
Fact-check & hallucination buster
I have generated an output about [topic]. Help me build a verification checklist to evaluate this text for hallucinations and false patterns. Spell out which primary sources, institutional policies, or datasets I should check to confirm dates, compliance terms, and numbers. Do not validate the text yourself, give me the human verification strategy. Text: [paste output]
Example: I generated a summary of the campus housing registration cycle. Build me a verification checklist and tell me which sources to check for dates and policy terms. Text: [paste draft]
Good for: Catching incorrect or misleading information, reviewing policy summaries, verifying numbers and dates
Why it's safe: Positions Copilot as an assistant and keeps verification in your hands
Privacy guard & record de-identifier
I need to draft a [document type] about a sensitive situation involving [general context]. Do not ask for or include names, ID numbers, exact dates, financial figures, or private records. Give me a fully de-identified, generalized template I can complete offline.
Example: I need to draft a communication about a student who has missed multiple classes and needs academic outreach. Give me a de-identified template with no names, IDs, or dates.
Good for: Building reusable templates for sensitive topics without exposing real records
Why it's safe: Helps prevent sharing personal details or protected information and gives you a structure to fill in yourself
Bias & inclusion audit partner
Review the text below for implicit bias, stereotypes, missing perspectives, language that excludes, accessibility barriers, or hidden assumptions about race, gender, disability, nationality, age, or socioeconomic status. Suggest specific, evidence-based revisions to make it more inclusive. Text: [paste text]
Example: Review the intake form below for bias and accessibility barriers for incoming first-generation students, and suggest inclusive revisions. Text: [paste form]
Good for: Improving public messaging, forms, and teaching materials
Why it's safe: Surfaces hidden tone and accessibility barriers early while you decide which revisions to accept
For Students — Enhancing learning & academic integrity
Use Copilot to study harder, not to cut corners. These prompts keep authorship and understanding with you.
Tutor mode: learn, don't shortcut
Act as a tutor for [course/topic]. Help me understand [concept/problem]. Don't give me the final answer right away. Ask me guiding questions, explain each step, and point out where my reasoning may be weak.
Example: Act as a tutor for Introductory Statistics. Help me understand what a null hypothesis is and when to use it. Don't give me the final answer right away, ask me guiding questions and point out where my reasoning is weak.
Good for: Homework help, active studying, concept breakdown, exam prep
Why it's safe: Supports active learning instead of doing the work for you
Concept explainer in three styles
Explain the core idea of [concept] three different ways: in plain language, with a real-world analogy, and with a rigorous technical explanation. Don't give me answers to graded work. Then give me two practice questions to test my understanding.
Example: Explain the core idea of molecular chirality three ways — plain language, an analogy, and a technical version — then give me two practice questions.
Good for: Deep comprehension and studying a tough concept from several angles
Why it's safe: Focuses on helping you understand the material rather than completing an assignment for you
Study guide builder
Create a study guide for [topic/chapter/unit]. Focus on the most important concepts, common misunderstandings, key vocabulary, and practice questions. Do not invent facts. Flag anything I should verify with my textbook, instructor, or course materials.
Example: Create a study guide for The French Revolution. Focus on key concepts, common misunderstandings, vocabulary, and practice questions, and flag anything I should verify.
Good for: Exam review, reading comprehension, structuring notes
Why it's safe: Builds a learning framework while reminding you to ground everything in official course materials
Draft feedback without rewriting
Review my draft below for clarity, organization, grammar, and argument strength. Do not rewrite the assignment for me. Give feedback in bullet points and suggest specific areas I can improve myself. Draft: [paste draft]
Example: Review my draft below for clarity, organization, grammar, and argument strength. Do not rewrite it, give bullet-point feedback. Draft: [500-word response on the impact of microplastics on marine biology]
Good for: Essays, discussion posts, lab reports, scholarship applications
Why it's safe: Keeps authorship fully with you and gives developmental feedback rather than replacement text
Citation & source reality check
I'm researching [topic]. Suggest search terms, possible databases, and the types of credible sources I should look for. Do not create citations unless you clearly mark them as needing verification. Warn me about areas where fake or unreliable sources are common.
Example: I'm researching the psychological effects of remote work on Gen Z employees. Suggest search terms, databases, and credible source types, and warn me where fake sources are common.
Good for: Research papers, capstone projects, building library-search skills
Why it's safe: Avoids AI-invented citations and strengthens your independent research strategy
Academic integrity self-check
Based on the course syllabus or assignment instructions below, help me identify which uses of AI would likely be acceptable, questionable, or not allowed. Do not assume permission. If AI use is allowed, help me write a short disclosure statement. Instructions: [paste instructions]
Example: Based on the instructions below, identify which AI uses are acceptable, questionable, or not allowed, and help me draft a disclosure statement. Instructions: AI may be used for brainstorming, but all writing must be your own work.
Good for: Using AI responsibly and staying within course rules
Why it's safe: Points back to course policy instead of guessing and helps you disclose AI use when required
Built-in Copilot tool for learning
In addition to open-ended chat, Copilot Chat includes a Study and Learn tool that WSU students may find helpful.
Study and Learn agent for Students
Study and Learn is a student-focused tool built on learning science principles. It helps students understand concepts, practice through activities like flashcards and matching, and prepare for exams using study guides and quizzes. Rather than simply providing answers, it is designed to foster reflective, critical thinking.
Find the Study and Learn agent in the Agent store
For Faculty — Policy, assessment & instruction
Develop clear policies, more equitable assessments, and richer materials while keeping student data out of your prompts.
AI-use policy builder (syllabus or assignment)
Draft clear, student-friendly AI-use guidelines for my [syllabus / specific assignment]. Separate uses that are allowed, allowed with disclosure, and not allowed. Address whether disclosure is required and how academic-integrity rules apply. Keep the language plain and supportive. Details: [paste course or assignment description]
Example: Draft student-friendly AI-use guidelines for my Freshman Composition argumentative essay on local climate policy. Separate allowed, allowed-with-disclosure, and not-allowed uses.
Good for: Syllabus language, assignment sheets, LMS overview pages
Why it's safe: Sets unambiguous expectations before students begin to reduce confusion and potential disputes
Rubric builder with bias check
Create a grading rubric for [assignment/task] with criteria, performance levels, and concise descriptions. Then review the rubric for possible bias, unclear expectations, or assumptions that could disadvantage first-generation, multilingual, disabled, or nontraditional students.
Example: Create a grading rubric for an undergraduate group presentation on renewable energy systems, then review it for bias and unclear expectations.
Good for: Rubric development and equitable assessment design
Why it's safe: Builds an equity review into the design process from the start
Assignment & rubric equity (UDL) auditor
Review the assignment or rubric below. Evaluate it for hidden assumptions, accessibility barriers, or phrasing that could disadvantage first-generation, disabled, or multilingual students. Suggest alternate explanation styles or Universal Design for Learning (UDL) modifications to make it more equitable. Text: [paste text]
Example: Review the group-presentation guidelines below for hidden assumptions and accessibility barriers, and suggest UDL modifications. Text: [paste guidelines]
Good for: Auditing existing assessments for equity and accessibility
Why it's safe: Runs an equity check on real materials without using or exposing student data
Lesson plan with multiple explanation styles
Create a lesson plan for teaching [concept] to [student level]. Include a brief explanation, an example, a class activity, discussion questions, and one misconception check. Provide at least three ways to explain the concept: plain language, an analogy, and a technical version.
Example: Create a lesson plan for teaching the basics of blockchain to undergraduate business majors, with an activity, discussion questions, a misconception check, and three explanation styles.
Good for: Lectures, workshops, lab instructions, recitations
Why it's safe: Provides structured, multi-modal explanations and built-in checks for understanding, making the content accessible for diverse learners without assuming prior knowledge or a single learning style
Student feedback templates (no private data)
Generate supportive, reusable feedback templates for students completing a [type of assignment]. Provide versions for strong work, developing work, and incomplete work, and point students toward office hours and tutoring. Do not reference individual names, grades, disability status, or personal circumstances.
Example: Generate supportive, reusable feedback templates for students completing a python data-cleaning assignment. Provide versions for strong work, developing work, and incomplete work, and point students toward office hours and tutoring. Do not reference individual names, grades, disability status, or personal circumstances.
Good for: Reusable LMS comment banks and faster grading with empathy
Why it's safe: Speeds up feedback while keeping individual student records out of the prompt
Hallucination-resistant teaching content
Using only the source text below, create a classroom handout on [topic]. Do not add outside facts. Separate “confirmed from the text” from “possible discussion questions,” and flag anything I should verify before sharing it with students. Source: [paste text]
Example: Using only the text below, create a classroom handout on the campus sustainability charter. Separate confirmed facts from discussion questions and flag anything to verify. Source: [3-page charter]
Good for: Assigned readings, journal articles, complex policy excerpts
Why it's safe: Restricts Copilot to your source, ensuring it does not confidently introduce unsupported material
Built-in Copilot tool for teaching
In addition to open-ended chat, Copilot Chat includes a teaching-focused tool that WSU instructors may find helpful.
Teach for Educators
Teach helps instructors plan and build instructional materials, lesson plans, quizzes, rubrics, and flashcards in one place within Copilot Chat. This can help free up more time to working directly with students.
For Staff — Streamlining operations & clarity
Communicate clearly and improve processes while keeping facts accurate and leaving compliance decisions to human review.
Plain-language policy explanation
Using only the official policy excerpt below, draft a plain-language explanation for [students / faculty / staff]. Do not add requirements, deadlines, exceptions, or contact information that aren't in the text. Include a short section titled “When to contact the office.” Excerpt: [paste policy]
Example: Using only the excerpt below, draft a plain-language explanation of the revised travel-reimbursement guidelines for faculty, with a “When to contact the office” section. Excerpt: [paste policy]
Good for: Office FAQs, website copy, help-desk responses
Why it's safe: Tightly controls the output, reducing invented policy details or false timelines
Email draft with fact lock
Draft a professional email to [audience] about [topic]. Use only the facts listed below. Do not invent dates, names, links, policy details, or next steps. Keep the tone clear, helpful, and concise. Facts: [paste facts]
Example: Draft an email to campus residents about scheduled power maintenance. Facts: Saturday 2–6 a.m.; affects North Quad; elevators offline; emergency lighting stays on.
Good for: Campus announcements, deadline reminders, operational notices
Why it's safe: Stops AI from improvising links or contacts, ensuring operational details stay accurate
Process improvement review
Review the administrative process below for clarity, unnecessary steps, bottlenecks, accessibility issues, and points where people may get confused. Suggest improvements, but do not recommend changes that affect compliance, eligibility, privacy, or institutional policy without human review. Process: [paste process]
Example: Review the steps below for the student parking-permit application and sticker pickup. Flag bottlenecks and accessibility issues, but leave compliance rules for human review. Process: [paste process]
Good for: Intake forms, internal workflows, service design, onboarding
Why it's safe: Offers efficiency ideas without allowing AI to override legal or compliance rules
Bias & tone review for student-facing messages
Review this message for tone, clarity, inclusiveness, accessibility, and unintended blame. Consider how it may land for first-generation, international, disabled, working, and stressed students. Suggest accurate, supportive revisions. Message: [paste message]
Example: Review this message for tone, clarity, inclusiveness, accessibility, and unintended blame. Consider how it may land for first-generation, international, disabled, working, and stressed students. Suggest accurate, supportive revisions. Message: 'Failure to submit your tax document by the deadline results in immediate cancellation of your aid.'
Good for: Registrar notices, financial-aid alerts, advising outreach
Why it's safe: Flags harsh or confusing administrative language to keep a supportive tone
Meeting notes into action items
Turn these meeting notes into a clear summary with decisions, action items, owners, and deadlines. Do not add tasks or decisions that aren't supported by the notes. Flag anything ambiguous or missing. Notes: [paste notes]
Example: Turn these notes from a department committee meeting into a summary with decisions, action items, owners, and deadlines. Do not add tasks or decisions that aren't supported by the notes. Flag anything ambiguous or missing. Notes: [paste notes]
Good for: Project coordination, committee work, operations tracking
Why it's safe: Acts as an organizer, cleanly separating what people agreed to from AI guesses
For Researchers — Strategy & integrity
Strengthen research design and writing while avoiding fabricated citations or unvalidated conclusions.
Literature search strategy
I'm researching [topic/question]. Help me develop a literature search strategy: keywords, related terms, possible databases, Boolean operators, inclusion/exclusion criteria, and questions to ask while reading. Do not invent citations. Clearly label anything that must be verified.
Example: I'm researching the efficacy of mRNA vectors in targeted tumor therapy for solid-tumor models. Help me develop a literature search strategy: keywords, related terms, possible databases, Boolean operators, inclusion/exclusion criteria, and questions to ask while reading. Do not invent citations. Clearly label anything that must be verified.
Good for: Systematic reviews, dissertation methodology, grant preparation
Why it's safe: Guides you toward real sources and search strategy without generating fake references
Research design critique
Critique this research design for validity, feasibility, bias, ethics, privacy, and possible confounding variables. Suggest improvements and identify what would need IRB, compliance, sponsor, or data-governance review. Study description: [paste description]
Example: Critique this design tracking adolescent social-media use and cortisol over six months. Identify validity threats and what needs IRB or data-governance review. Description: [paste description]
Good for: Early study planning, methods peer review, draft grant narratives
Why it's safe: Surfaces design, privacy, and ethics risks early, giving you prompts for your IRB or compliance submission
De-identified data analysis planning
I have a de-identified dataset about [general topic]. Help me think through possible analysis approaches, assumptions to check, limitations, and visualization ideas. Do not ask for or require identifiable personal data. Do not draw conclusions without validated results.
Example: I have a de-identified dataset of undergraduate mental-health surveys across three campuses. Help me plan analysis approaches, assumptions, limitations, and visualizations. Do not ask for or require identifiable personal data. Do not draw conclusions without validated results.
Good for: Exploratory analysis planning and structuring visualization ideas
Why it's safe: Keeps identifiable data out of your prompt and avoids premature, unvalidated conclusions
Overclaiming & research integrity review
Review this abstract or draft section for overclaiming, unsupported causal language, vague claims, missing limitations, and statements that need citations. Do not add new findings. Suggest more precise wording. Text: [paste text]
Example: Review this draft abstract for overclaiming, unsupported causal language, vague claims, missing limitations, and statements that need citations. Do not add new findings. Suggest more precise, measured wording. Text: 'Our minor curriculum change definitively solves the retention crisis permanently.’
Good for: Manuscript editing, abstract polish, poster text, grant progress narratives
Why it's safe: Enforces measured, honest claims that stay fully supported by your evidence
Code support with validation steps
Help me debug or improve this code for [purpose]. Explain what each suggested change does. Don't assume the code is correct. Include validation checks, edge cases, and ways to confirm the output is accurate. Code: [paste code]
Example: Help me debug this Python/pandas script for cleaning a messy CSV with missing dates. Don't assume the code is correct. Explain each change and include validation checks and edge cases, and ways to confirm the output is accurate. Code: [paste script]
Good for: Data-cleaning scripts, pipelines, reproducible research workflows
Why it's safe: Shifts focus from blind code acceptance to testing, edge cases, and reproducibility
Universal safety add-ons
These short instructions can be added to the end of most Copilot prompt to support privacy and accuracy. Use one or more as needed.
Reduce hallucinations
Do not invent facts, sources, dates, policies, names, or statistics. If you're uncertain, say so and tell me what a human should verify.
Force source discipline
Use only the information I provide below. Do not use outside knowledge or training data unless I explicitly ask for it.
Protect privacy
Do not ask me to provide names, ID numbers, private records, health information, grades, financial data, or other sensitive personal information.
Check bias & inclusivity
Review your response for implicit bias, stereotypes, missing perspectives, accessibility barriers, and assumptions about background, ability, identity, or socioeconomic status.
Keep AI from making decisions
Do not make a final decision or recommendation about a person's grade, eligibility, employment, discipline, admission, funding, accommodation, or status. Provide only a structured framework for human review.
Make the output reviewable
Break your response into four parts: confirmed information from my text, assumptions, potential risks, and items requiring human review.
Improve transparency
Write a brief, professional AI-use disclosure statement describing how this output was generated and what human review is still needed.
Preserve confidentiality
Treat this as confidential. Do not reuse this text for any other purpose, and process no unpublished manuscript, finding, or grant narrative beyond answering my immediate request.