Using AI Responsibly in Your Ontario Classroom
The Ontario guide to Artificial Intelligence in teaching and learning — what AI should and should not do, how to protect student privacy, how to keep professional judgement at the centre of every decision, and twenty-five responsible uses you can start this week.
- Published
- August 23, 2026
- Reading time
- 30 minutes
- Grades
- Kindergarten to Grade 8
- Subjects
- All subjects
Quick answer
What does responsible AI use look like in an Ontario classroom?
Responsible AI use in Ontario classrooms means AI supports rather than replaces professional judgement. Teachers use it to reduce administrative workload, generate drafts and ideas, and personalise learning — never to make final assessment decisions. No identifiable student information is entered. Every output is verified against the Ontario Curriculum and Growing Success, and the teacher remains accountable for every instructional decision.
- AI drafts; teachers decide. Evaluation and reporting remain the teacher's professional judgement under Growing Success (2010)
- No personally identifiable student information should ever be entered into an AI tool
- Every AI output must be checked against the Ontario Curriculum before it reaches students
- AI is most valuable on repetitive tasks — first drafts, reformatting, differentiation variants, administrative writing
- Teaching students to use AI critically is now part of digital citizenship, not an optional extra
A staff room in late August, two coffees and one uncomfortable question
A few years ago, the staff room argument was about smartphones. Should students be allowed to have them out? In the bag? In a caddy by the door? Everyone had a position, and everyone was certain. This August the argument has changed. A Grade 7 teacher says she has already had three students hand in writing that was clearly not theirs. A Grade 2 teacher says she used AI to draft her welcome letter on Sunday and finished it in six minutes instead of forty. A department head says the board is drafting guidelines but nobody has seen them. And a first-year teacher says almost nothing, because she has been using AI to help plan since her practicum and is not sure whether that is something to admit.
Some educators are excited. Others are cautious. Most Ontario teachers feel both at the same time, on the same day. The honest reality is that Artificial Intelligence is already part of education — in students' pockets, in board procurement conversations, in the tools embedded in software teachers already use. The question is no longer whether you will encounter AI. It is how to use it responsibly, ethically, and in ways that genuinely improve learning rather than quietly hollowing it out. That is what this guide is for.
Key Takeaways
Everything below in under a minute
- You are still the expert. AI produces plausible text; only you know whether it is right for this curriculum, this class and this child.
- AI should reduce workload, not thinking. If a task requires knowing your students, do it yourself. If it requires typing something you have typed a hundred times, delegate it.
- Student privacy comes first, always. Describe situations generically, use initials or 'a Grade 5 student', and never paste identifiable information into any AI tool.
- Verify everything. Treat every AI output as a confident first draft from a keen student teacher who has never read the Ontario Curriculum.
- Use AI to personalise, not to standardise. Its highest classroom value is producing four versions of a task instead of one.
- Teach students to use AI well. Fact-checking, citation, prompt-writing and knowing when not to use it are teachable skills at every grade.
- Design assessment around process, not just product. Authentic tasks, conversations, observations and revision make integrity a design question rather than a policing question.
The Responsible AI Framework for Ontario classrooms
Six checks, applied in order, before any AI output reaches a student, a family or a report card. They take about ninety seconds once they become habit, and they are what separates professional use from careless use.
Purpose
Is this task genuinely repetitive or generative — or is it a decision that requires knowing this student? Only the first belongs to AI.
Privacy
Have I removed every identifying detail? Names, initials that identify, diagnoses, addresses, incidents that a reader could trace to one child.
Prompt
Have I given the grade, subject, Ontario curriculum context, the purpose and the constraints? Vague prompts produce generic output that costs more time to repair.
Proof
Is the output accurate, curriculum-aligned, developmentally appropriate, bias-free and written in Canadian English? Verify before you use it.
Professional judgement
Would I make this decision the same way without the AI? If the tool changed my judgement rather than my typing, stop and reconsider.
Transparency
Could I explain to a parent, a colleague or my principal exactly how this was produced? If the answer is no, change how you produced it.
Purpose, privacy, prompt, proof, professional judgement, transparency. Six checks that keep AI in the assistant's chair where it belongs.
Section 1 — Why AI matters in Ontario education
Ontario teachers are not short of ideas. They are short of hours. The average elementary teacher in this province plans across five or more subjects, differentiates for a class that may include several students with IEPs and several multilingual learners, collects triangulated evidence across three terms, writes hundreds of report card comments, and communicates with dozens of families — while teaching all day. Artificial Intelligence has arrived into that reality, and it matters for one honest reason: a meaningful share of a teacher's week is spent on tasks that require professional knowledge but not professional presence. Reformatting a plan. Writing the fourth version of the same instruction at a different reading level. Turning three observation notes into a sentence a parent will understand. That work has to be done well, but it does not have to be done from scratch.
The change is faster than the guidance
It is worth naming the discomfort directly. The technology has moved faster than policy, faster than professional learning, and faster than most board procedures. Teachers are being asked to make judgement calls in a space where the guardrails are still being built. That is genuinely uncomfortable, and the anxiety many educators feel is not technophobia — it is professionalism. The right response is not to wait for perfect guidance, nor to adopt everything uncritically. It is to adopt a defensible personal standard now: privacy protected, outputs verified, judgement retained, use transparent. If you can meet those four conditions, you are on solid ground regardless of which guidance arrives next. Keep an eye on Ontario Education Updates for policy developments as they are published.
Students are already there
By the time a policy conversation reaches a staff meeting, the students have usually settled the practical question themselves. Intermediate students in particular are using AI for homework, for summarising, for translation, for study questions and, yes, sometimes for writing that should have been theirs. Pretending otherwise does not protect learning; it just removes the teacher from the conversation. The most protective thing an Ontario teacher can do is to be the adult who taught them how to use it well — how to check a claim, how to cite a tool, how to tell the difference between using AI to think better and using it to avoid thinking. That is digital literacy, and it belongs in every classroom, not only in Grade 8.
Section 2 — What AI should and should not do
The clearest way to think about this is to sort tasks by what they require. Some tasks require language production: drafting, rephrasing, formatting, generating variants, summarising, brainstorming. AI is genuinely good at these, and delegating them costs nothing pedagogically. Other tasks require knowledge of a specific child, professional accountability, or a judgement that a parent might one day ask you to justify. AI must not touch those. The dividing question is simple: does this task require knowing the student? If yes, it is yours. If no, it is a candidate for delegation, subject to the six checks in the framework above.
What AI should help with
Lesson planning drafts and structures. Differentiation variants at multiple entry points. Brainstorming — hooks, provocations, consolidation questions, misconception probes. Rubric drafts across the four Achievement Chart categories. Success criteria written in student-friendly language. First drafts of descriptive feedback based on evidence you supply. Parent communication drafts you then personalise. Assessment ideas and question banks. Organisational work: agendas, checklists, timelines, supply teacher plans, newsletter scaffolds. In every case, AI is doing the typing and the structuring. You are doing the deciding.
What AI should never replace
Professional judgement, which is the legal and ethical core of Ontario assessment. Relationships with students and families, which cannot be automated and are the strongest predictor of how a difficult year goes. Assessment and evaluation decisions — Growing Success is explicit that evaluation is the teacher's professional judgement about the quality of student work against curriculum expectations. Human empathy, particularly in the moments that matter most. Knowledge of individual students, which is accumulated in a thousand small observations no tool has access to. And final evaluation, which is yours: the level, the grade, the comment that goes on the provincial report card, and your ability to explain all three.
Section 3 — Protecting student privacy
This is the non-negotiable section. Ontario educators operate under strong privacy expectations: municipal and provincial freedom of information legislation governs school board records, the Education Act governs the Ontario Student Record, and PIPEDA governs commercial organisations handling personal information in the course of commercial activity. The practical rule that keeps you safe under all of them is straightforward: personally identifiable student information should never be entered into a general-purpose AI tool. Not names. Not initials that identify within a small community. Not diagnoses, safety plans, custody arrangements, immigration status, medical details, or incident narratives detailed enough to identify a child.
Safe prompting in practice
Safe prompting is a skill, and it takes about a week to become automatic. Instead of 'Write a report card comment for Aiden Mercier, who has ADHD and struggles to finish work', write 'Write a Grade 4 mathematics report card comment for a student who demonstrates strong number sense orally, needs support to complete written work independently, and benefits from chunked tasks.' The second prompt contains everything the tool needs and nothing that identifies a child. Notice that it is also a better prompt: it is specific about evidence rather than about diagnosis, which is exactly how Growing Success asks us to write about students anyway.
Know where the data goes
Not all AI tools are equivalent, and 'free' usually means the data is doing something. Before adopting a tool for regular professional use, ask four questions: where is the data stored and processed, is it used to train models, who can access it, and has my board approved it? Education-specific platforms built for a jurisdiction generally answer these questions in writing; general consumer chatbots often do not. Ontario Curriculum AI was built for this context — Canadian hosting, encryption in transit and at rest, no selling of data, and interfaces that actively discourage entering identifiable student information. Our full posture is documented on the security page and in the trust and compliance overview. Whichever tool you choose, insist on being able to answer those four questions.
Section 4 — Responsible AI for lesson planning
Planning is where AI pays for itself fastest, because most of the time cost in planning is not deciding what to teach — it is writing it down in a usable form. You already know your class needs another day on fractions before moving to comparison. What takes forty minutes is turning that into a three-part lesson with a minds-on hook, a task, consolidation questions, success criteria, differentiation and an exit ticket. That is structural work, and it can be drafted in a minute so that your forty minutes goes to the parts that need you.
A responsible planning workflow
Start with the curriculum, not the tool. Choose the overall and specific expectations you are actually assessing, then give the AI that context: grade, subject, expectation, class realities, the time you have and the resources available. Ask for a three-part lesson. Read the draft critically and expect to reject part of it — AI tends to over-produce activities and under-produce consolidation. Rewrite the learning goal in your own voice. Check that the success criteria describe quality rather than compliance ('explains thinking using a model' rather than 'completes all questions'). Then adjust for the students you have. The whole workflow is set out in detail in the Ontario Lesson Planning Workflow, and the planning tools live in the lesson planning hub.
Prompting well: the difference in practice
Weak prompt: 'Write a Grade 5 math lesson on fractions.' You will get a generic American-flavoured lesson you then have to rebuild. Strong prompt: 'Write a three-part Ontario Grade 5 mathematics lesson on comparing fractions with unlike denominators. Class of 26, four students working at a Grade 3 level, two multilingual learners in early stages of English. Fifty minutes. Include a minds-on estimation task, a partner task using number lines, three consolidation questions, student-friendly success criteria, and an exit ticket. Canadian English and Ontario terminology.' The second prompt produces something you can actually adapt, and it took eleven extra seconds to write.
Section 5 — Responsible AI for assessment
Assessment is where the line between support and substitution needs to be brightest. AI can legitimately help with the design and the wording of assessment: drafting rubrics across the four Achievement Chart categories, generating diagnostic tasks, producing exit tickets aligned to an expectation, suggesting observation look-fors, and turning your evidence into descriptive feedback that names a strength, a criterion and a next step. What it must never do is determine achievement. The level is a professional judgement based on evidence of learning gathered over time through observations, conversations and student products — and that judgement belongs to the teacher who gathered it.
Feedback is the highest-value use
If you use AI for one assessment purpose, make it descriptive feedback. Research on feedback is consistent: it works when it is specific, timely, and tells the student what to do next. It fails when it is late or generic. The practical barrier for Ontario teachers has never been knowing that — it is that writing twenty-six pieces of specific feedback takes an evening. Supply the evidence yourself ('used a number line accurately but did not explain why the comparison worked'), and let the tool draft the wording. You will produce better feedback for more students, faster, and every word of it originates in evidence you observed. The student feedback generator is built around exactly this pattern.
Section 6 — Helping students use AI responsibly
Students need to be taught this the way they were taught to evaluate a website: explicitly, repeatedly, and at an age-appropriate level. The competencies are clear enough to plan for. Knowing what AI is and what it is not. Understanding that it can be confidently wrong. Fact-checking a claim against a reliable source. Citing tool use honestly. Writing a useful prompt. Knowing which tasks are their own thinking and which are legitimate assistance. And recognising bias — that a tool trained on the internet will reproduce what the internet contains, including whose stories get told and whose do not.
A three-lesson mini-unit that works in any junior or intermediate class
Lesson one: what AI actually does. Show that it predicts text. Ask it something the class knows well and find the error together. Build the 'confidently wrong' anchor chart. Lesson two: prompting and checking. Students write prompts, compare outputs, then verify one factual claim each using a source from your library or the curriculum. Lesson three: ethical use. Co-construct the class agreement — what AI may help with (brainstorming, checking spelling, explaining a concept a different way) and what must be their own (their ideas, their draft, their thinking), and how to acknowledge use. Post the agreement. Refer to it all year.
Section 7 — AI and academic integrity
The fear-based version of this conversation is unproductive and, worse, unnecessary. Detection tools are unreliable and produce false accusations that damage relationships, particularly for multilingual learners whose writing patterns are misread. The productive version is a design conversation. If a task can be completed in one sitting, with no visible process, and assessed only on the finished artefact, it was always vulnerable — to copying, to over-helpful parents, to purchased essays and now to generative tools. Designing for process makes the question mostly disappear.
What integrity-resilient assessment looks like
Drafts that are seen and dated, so the process is visible. Conferences where students explain their thinking, which take four minutes and tell you more than any detector. Oral components — explain your solution, defend your interpretation, teach it to a partner. Tasks anchored in local, current or personal context that a general model cannot produce convincingly. Revision cycles where the growth between drafts is what is assessed. In-class writing for at least some tasks. Performance tasks and demonstrations. Reflection on the process itself, which is also excellent Learning Skills evidence. None of this is new pedagogy; it is Ontario assessment done well, and Ontario assessment strategies has more on building it into a unit.
Section 8 — Common AI myths in Ontario staff rooms
Five claims come up in almost every conversation, and each contains a partial truth that deserves an honest answer rather than a slogan. Taking them seriously is what makes a staff conversation productive.
'AI will replace teachers'
It will not, and the reason is not sentimental. Teaching is a relational and judgement-intensive profession: noticing that a normally chatty student has gone quiet, deciding that today's plan should be abandoned, knowing that this family needs a phone call rather than an email, holding a standard while communicating belief in a child. None of that is a language problem. What AI will change is the composition of a teacher's week, by absorbing tasks that were never the reason anyone entered the profession. The teachers who thrive will be the ones who use that time on the relational and judgement work that no tool can do.
'AI writes perfect lessons'
It writes plausible lessons. There is a large difference. AI output typically over-produces activities, under-produces consolidation, drifts toward American terminology, and cannot know that your Tuesday afternoon block follows gym or that six of your students need the vocabulary pre-taught. A first draft that is seventy per cent right is genuinely valuable — but the thirty per cent is the teaching, and it is yours.
'Students should never use AI'
Students will encounter AI in postsecondary study, in apprenticeships and in nearly every workplace. A blanket prohibition removes the teacher from a formative conversation without removing the technology. The defensible position is task-specific: some tasks are for AI-assisted work with acknowledgement, and some are protected because the thinking is the point. Say which is which, explicitly, for each task. Students respect a clear line far more than a vague one.
'AI is always accurate'
It is fluent, which reads as accurate. It fabricates references, misstates curriculum expectations, invents Ontario policy details, and produces statistics with total confidence. Anything factual, curricular or policy-related must be verified against the source — the Ontario Curriculum, Growing Success, or your board's own documents. Treat fluency as a presentation format, not as evidence.
'AI removes professional judgement'
Only if you let it. A calculator did not remove mathematical reasoning from teaching; it moved the reasoning to a different place. AI drafts wording; you decide substance. The risk is real but specific: it is the drift that happens when a teacher is exhausted in late June and accepts an output because reading it carefully feels like more work. Guard against that with a rule rather than willpower — never send, submit or assign an AI-assisted output you have not read end to end.
Section 9 — Practical classroom examples
The examples below are realistic workflows rather than demonstrations. Each names the teacher's goal, where AI genuinely helps, and — most importantly — where professional judgement remains essential. The pattern is the same at every division: AI handles the production, the teacher handles the decisions.
Section 10 — Building an AI-ready classroom
An AI-ready classroom is not a classroom full of devices. It is a classroom where expectations are explicit, technology use is purposeful, students think critically about the tools they use, and the teacher models responsible practice openly. Four things make it work. First, clear task-level expectations: for each significant task, say whether AI assistance is permitted, in what way, and how it should be acknowledged. Second, co-constructed agreements, which students follow far better than imposed rules. Third, transparency with families — a short paragraph in your welcome letter or newsletter explaining that AI supports your planning and that student privacy is protected does enormous work. Fourth, alignment with your board's policy, checked before you scale anything.
Talking to families about AI
Families ask two questions: is my child's information safe, and is a computer marking my child's work? Answer both plainly. Your information is protected — no identifiable student information is entered into any AI tool. And no: your child's teacher determines every level and every comment based on evidence gathered in the classroom, and uses AI only to help word and organise professional work. That is a two-sentence answer that resolves nearly every concern. The parent communication templates guide has wording you can adapt for a newsletter or welcome letter.
The strongest argument for responsible AI in Ontario classrooms has nothing to do with technology. It is that a teacher who finishes report cards in three evenings instead of eight is a better teacher in January — more present, more patient, more available to the student who needs ten minutes. Efficiency is not the goal. Capacity for the human work is the goal.
Ontario classroom examples
The same idea, written the way it actually sounds in each division.
Teacher goal: build a documentation panel and pedagogical documentation notes for an inquiry into the schoolyard that emerged from children's questions about puddles. The educator team wants provocations, vocabulary, and family-friendly wording for the panel — without spending an evening writing it.
How AI helps: drafts five provocations connected to the children's puddle questions, suggests inquiry vocabulary, and turns the educator team's observation notes into a short, plain-language panel caption for families. Where judgement remains essential: which of the children's questions is actually worth pursuing, what the documentation reveals about individual children's thinking, and every frame of reference statement — those come from educators who were standing in the puddle with them.
The Kindergarten Program (2016), Belonging and Contributing; Demonstrating Literacy and Mathematics Behaviours
Teacher goal: run a narrative writing unit with clear success criteria and feedback for twenty-three students, six of whom need significant scaffolding and two of whom are early-stage multilingual learners.
How AI helps: drafts student-friendly success criteria for a Grade 2 narrative, generates a sentence-starter scaffold and a picture-supported planning organiser, and turns the teacher's conferencing notes ('strong beginning, events out of order, no ending') into warm descriptive feedback a seven-year-old can act on. Where judgement remains essential: deciding that this class needs two more days on sequencing before drafting, choosing which children get which scaffold, and every judgement about how a child is progressing.
Ontario Language curriculum (2023), Strand D: Composing and Creating
Teacher goal: design a diagnostic for fractions in September, then differentiate a comparison task four ways for a class with a five-year spread in demonstrated understanding.
How AI helps: drafts a ten-question diagnostic with a range of entry points, produces four versions of the comparison task (scaffolded with number lines, core, extension with reasoning, and vocabulary-supported), and suggests three misconception probes. Where judgement remains essential: interpreting the diagnostic results against what the teacher already knows about these students, deciding who gets which version — a decision about confidence and readiness as much as skill — and determining every achievement level.
Ontario Mathematics curriculum (2020), Strand B: Number
Teacher goal: build a culminating task on biodiversity that is engaging, assessable across all four Achievement Chart categories, and resistant to being completed by a chatbot the night before.
How AI helps: drafts a rubric across Knowledge and Understanding, Thinking, Communication and Application; suggests local Ontario biodiversity contexts; and generates reflection prompts for the process log. Where judgement remains essential: anchoring the task in the students' own schoolyard survey data, which no model can fabricate; the four-minute conference where each student explains their reasoning; and the professional judgement that produces the final level from the whole body of evidence.
Ontario Science and Technology curriculum (2022), Strand B: Life Systems
Teacher goal: teach responsible AI use explicitly while running a historical inquiry, and design the task so that student thinking is the assessed object.
How AI helps: generates a deliberately flawed sample AI response for students to critique, drafts the class AI-use agreement template, and produces differentiated research question stems. Where judgement remains essential: the teaching of critical evaluation itself, the in-class writing and conferencing that show thinking, and the decision about how AI acknowledgement is treated in this class — which the teacher makes explicit on every task sheet rather than leaving to inference.
Ontario Language curriculum (2023), Strand B: Critical Literacy; Ontario Social Studies, History and Geography
Teacher goal: draft accommodations and measurable annual programme goals for a student with an IEP, written in professional language and consistent with what the team agreed at the meeting.
How AI helps: converts a generic description of need ('difficulty with sustained written output, strong oral expression, benefits from chunking') into professionally worded accommodations and measurable goal language. Where judgement remains essential: everything about the individual student. The description entered contains no identifying information; the team's knowledge of the child, the IPRC context, the family's input and the final wording are the professionals' responsibility, and every draft is reviewed against the actual student before it enters the IEP.
Regulation 181/98; Ontario IEP standards
See it in an Ontario classroom
The same idea at your grade level — teacher language, anchor chart, lesson flow and the evidence it leaves behind.
November in Ontario schools: Open Report Card Comments.
AI-assisted teaching vs AI-dependent teaching
The same tool, used two ways. The difference is not how much AI is used — it is whether professional judgement stays upstream of it.
Who decides
AI-assisted (responsible): Teacher decides, then uses AI to draft and word the decision
AI-dependent (risky): AI proposes, teacher accepts because it sounds reasonable
Student data
AI-assisted (responsible): Generic descriptions, no identifying details, ever
AI-dependent (risky): Names, diagnoses or incident details pasted in for convenience
Curriculum
AI-assisted (responsible): Expectations chosen first, output checked against them
AI-dependent (risky): Output assumed to be curriculum-aligned because it mentions the grade
Assessment
AI-assisted (responsible): Level determined from triangulated evidence; AI helps word the comment
AI-dependent (risky): Work pasted in and a level requested
Verification
AI-assisted (responsible): Every output read end to end before use
AI-dependent (risky): Output copied straight into a plan, comment or email
Students
AI-assisted (responsible): Explicitly taught when and how AI may be used, and why
AI-dependent (risky): Rules unstated; integrity handled by suspicion after the fact
Transparency
AI-assisted (responsible): Comfortable explaining the process to a parent or principal
AI-dependent (risky): Process is something to keep quiet about
Time saved
AI-assisted (responsible): Reinvested in feedback, conferring and relationships
AI-dependent (risky): Absorbed with no change to the work that matters
If a use of AI would fail the transparency row, it will usually fail one of the others as well. That row is a reliable single check when you are unsure.
How to implement this tomorrow
A four-week plan for a teacher who wants to use AI well without changing everything at once. Nothing here requires board approval beyond following your existing policy, and every week ends with something you can actually use.
Preparation
- Read your board's AI guidance if it exists; note anything more restrictive than this guide
- Choose one Ontario-specific tool rather than three general ones
- Write your personal privacy rule on a sticky note: no names, no diagnoses, no incidents
Materials
- One upcoming lesson you already teach
- One set of assessment evidence you have already collected
- Your class list of accommodations, described generically
- Fifteen minutes, once a week
Time required
Four weeks, about fifteen minutes each week
Week 1 — Add, do not replace
15 minutesTeacher actions
- Take a lesson you already teach and ask AI only for what you did not have time to build: one scaffold, one extension, one misconception probe
- Read the output critically and note what you rejected — this is how you learn the tool's failure modes
- Do not use AI for anything student-facing yet
Week 2 — Feedback from your own evidence
15 minutesTeacher actions
- Choose five students and write your evidence in one line each — strength, criterion, next step
- Draft descriptive feedback from that evidence, with no identifying information entered
- Edit every draft into your own voice before it reaches a student
Student actions
- Receive specific, actionable feedback within two days of the task
Week 3 — Teach students
One period plus 10 minutesTeacher actions
- Run one lesson on 'confidently wrong': show an AI output and find the errors together
- Co-construct a class AI-use agreement and post it
- Add an AI-use line to your next task sheet: not permitted, brainstorming only, or permitted with acknowledgement
Student actions
- Identify inaccuracies in an AI output
- Contribute to the class agreement about when AI helps thinking and when it replaces it
Week 4 — Systems and families
15 minutesTeacher actions
- Apply AI to one genuinely administrative job: a supply teacher plan, a newsletter, a meeting agenda
- Add two sentences to your newsletter explaining how AI supports your planning and how privacy is protected
- Decide what you will keep doing, and write it down as your personal standard
Reflection
- Which tasks did AI genuinely improve, and which did it only make faster?
- Did I ever accept an output I had not fully read? What made that moment tempting?
- Did any prompt I wrote contain information I would not want printed and left on a photocopier?
- Where did the time I saved actually go?
- What do my students now understand about AI that they did not in September?
Challenges and solutions
- Output sounds American or uses 'standards' language
- Name Ontario explicitly in the prompt — grade, curriculum, Canadian English — and use an Ontario-specific tool where the terminology is built in rather than requested.
- It takes longer to fix the output than to write it myself
- Almost always a prompt problem. Add grade, subject, expectation, class context, time available and constraints. If it is still faster to write it yourself, that task is not a good AI candidate — stop using it there.
- I am not sure whether my board permits this
- Ask, in writing, before you scale it. In the meantime apply the strictest reasonable standard: no identifiable information, no delegated judgement, and nothing student-facing that you have not fully reviewed.
- A colleague thinks using AI is cheating
- Draw the same line you draw for students: assistance with production is not substitution for thinking. Show them a draft you rejected — it is usually the most persuasive thing you can do.
- Students are using AI to avoid the work
- Treat it as a task design signal rather than a discipline problem. Add a visible draft, a four-minute conference, or a local context the model cannot fake.
- I feel uncomfortable and cannot say why
- Run the transparency check: could you explain this to a parent? Discomfort usually means the answer is no, and that instinct is professional judgement doing its job.
Checklist
The Ontario Responsible AI Classroom Pack
Everything from this guide in one printable set: the Responsible AI Framework, the should/should-not tables, the twenty-five responsible uses, the questions to ask before you prompt, and a class AI-use agreement template.
- The six-check Responsible AI Framework, one page
- Task-by-task 'should AI help?' decision table
- Teacher responsibility vs AI support chart for staff meetings
- 25 responsible uses of AI, organised by category
- The questions every teacher should ask before prompting
- A co-constructable class AI-use agreement for Grades 4–8
- Two-sentence family wording for a newsletter or welcome letter
Ontario terms in this article
Definitions, in Ministry language
- Differentiated Instruction
- Differentiated instruction adjusts content, process, product or learning environment in response to student readiness, interest and learning preference, while keeping the same curriculum expectations for the class. Differentiated Instruction resources
- Ontario Ministry of Education
- The Ontario Ministry of Education sets curriculum policy, assessment policy and reporting requirements for publicly funded schools in Ontario. Ontario Ministry of Education resources
- Professional Judgement
- Professional judgement is the informed decision-making Ontario teachers use to determine a grade from the whole body of evidence, weighting the most recent and most consistent work rather than averaging every mark. Professional Judgement resources
- Growing Success (2010)
- Growing Success (2010) is Ontario's policy document for assessment, evaluation and reporting in Grades 1 to 12. It defines the Achievement Chart, the learning skills and work habits, and the role of professional judgement in determining a report card grade. Growing Success (2010) resources
- Success Criteria
- Success criteria describe what successful attainment of a learning goal looks like. They are co-constructed with students where possible and used for descriptive feedback, self-assessment and peer assessment. Success Criteria resources
- Rubric
- An Ontario rubric describes achievement at Levels 1 to 4 against the four Achievement Chart categories, using the Ministry qualifiers limited, some, considerable and thorough. Rubric resources
Frequently Asked Questions
Decision table
Task by task: should AI help?
The dividing question is always the same — does this task require knowing the student? Print this and keep it beside your planning book.
Drafting a three-part lesson structure
YesStructural writing. You choose the expectations and adapt the draft.
Producing four differentiated versions of a task
YesProduction work. Matching students to versions stays with you.
Writing student-friendly success criteria
YesWording work. Check that criteria describe quality, not compliance.
Drafting a rubric across the four categories
YesDraft only. Verify the descriptors assess the expectation you chose.
Wording descriptive feedback from your evidence
YesYou supply strength, criterion and next step; AI supplies the sentence.
Drafting a message home about a concern
With careNo identifying details. Personalise before sending, and never send angry.
Writing a report card comment
With careOnly after you have determined the level from triangulated evidence.
Drafting IEP accommodation language
With careGeneric description of need only. The team reviews against the actual student.
Building a supply teacher plan
YesPurely administrative. Remove any student-specific medical or safety detail.
Determining a student's achievement level
NoEvaluation is professional judgement under Growing Success. Not delegable.
Deciding whether a student needs an intervention
NoRequires knowledge of the child, the family and the context.
Writing a safe schools or incident report
NoIdentifiable, sensitive and legally significant. Write it yourself.
Interpreting diagnostic results for your class
NoInterpretation depends on what you already know about these students.
Responding to a family in crisis
NoHuman empathy in real time. Pick up the phone.
Where the line sits
Teacher responsibility and AI support, side by side
A staff meeting handout. Nothing in the left-hand column can be delegated, and nothing in the right-hand column is worth an evening of your time.
Curriculum
You: Selects the overall and specific expectations and confirms alignment
AI: Drafts activities and tasks against the expectations you name
Planning
You: Decides pacing, sequence and what this class needs next
AI: Produces the lesson structure, scaffolds and extensions
Differentiation
You: Matches students to entry points and adjusts in the moment
AI: Generates multiple versions and language supports
Assessment design
You: Chooses what evidence is needed and how it will be gathered
AI: Drafts rubrics, exit tickets, diagnostics and look-fors
Evaluation
You: Determines the level using triangulated evidence — always
AI: Nothing. This is professional judgement
Feedback
You: Observes the learning and identifies the next step
AI: Words the feedback clearly and warmly from your evidence
Reporting
You: Decides what the comment must say and stands behind it
AI: Drafts Ontario-appropriate wording for your decision
Family communication
You: Knows the relationship, the history and the right tone
AI: Drafts and rewrites; you personalise and send
Student wellbeing
You: Notices, responds, refers and follows up
AI: Nothing
Privacy
You: Controls what is entered, always generically
AI: Nothing — it cannot protect what you type in
Privacy and ethics
Appropriate use and what to avoid
Seven pairs that cover almost every judgement call an Ontario teacher will face in a school year.
Appropriate
- Describing a need generically: 'a Grade 3 student who reads fluently but rarely infers'
- Using AI to draft a comment after you determined the level
- Generating three misconception probes for tomorrow
- Asking for a plain-language explanation of a curriculum idea
- Teaching students to critique AI output
- Drafting a newsletter or supply teacher plan
- Using AI to rewrite a draft you wrote while frustrated
Not appropriate
- Entering a student's name, initials, diagnosis or IEP document
- Pasting student work and asking what level it is
- Using a generated lesson without reading it
- Trusting an AI summary of Ontario policy without checking the source
- Using detection software to accuse a student
- Drafting a safe schools report or a custody-related message
- Letting AI decide what to say about a child
Feature
25 ways Ontario teachers can responsibly use AI
Every one of these keeps professional judgement with the teacher and identifiable student information out of the tool. Start with one from Planning and one from Assessment.
Planning
Draft a three-part lesson from chosen expectations
Give the grade, subject, expectation and time available, then rewrite the learning goal in your own voice.
Build a unit overview and pacing sequence
Ask for a five-week arc, then move things — you know which week has three assemblies in it.
Generate minds-on hooks and provocations
Ask for six, use one. Brainstorming is where AI has almost no downside.
Write student-friendly success criteria
Then check each one describes quality rather than task completion.
Draft learning goals in student language
'I am learning to…' phrasing, checked against the expectation you actually chose.
Produce consolidation questions
AI under-produces these on its own, so ask for them explicitly every time.
Plan a long-range overview for the year
A first draft you rearrange around reporting periods, trips and your own strengths.
Assessment
Draft a rubric across all four categories
Verify each descriptor assesses the expectation and not simply effort.
Generate a September diagnostic
Multiple entry points so the results tell you something about every student.
Create exit tickets aligned to today's goal
Two questions, ninety seconds, immediate evidence for tomorrow's grouping.
Draft observation look-fors
Turns a vague 'watch for understanding' into three things you can actually record.
Turn evidence into descriptive feedback
You supply the observation; AI supplies the wording students can act on.
Build conferencing question stems
Four questions that make thinking visible in a four-minute conversation.
Differentiation
Produce scaffolded, core and extension versions
The single biggest time saving available to an elementary teacher.
Generate vocabulary supports for multilingual learners
Key terms, plain definitions and visuals to pre-teach before the lesson.
Draft accommodation wording for an IEP
Generic description of need in; professional wording out; team review always.
Rewrite a text at three reading levels
Same content, different access. Check the vocabulary against your class.
Communication
Draft a welcome letter or curriculum night note
Six minutes instead of forty, then add the two personal details only you can.
Rewrite an email written while frustrated
Paste your draft, ask for a calmer professional version, then reread in the morning.
Write a weekly newsletter from four bullet points
Same four headings each week keeps it to eleven minutes.
Reporting
Word report card comments from your evidence
Level determined first by you; AI only expresses the judgement.
Draft Learning Skills comments for all six skills
E / G / S / N wording that describes observable behaviour.
Organisation and professional learning
Write a supply teacher plan in minutes
On the morning you wake up unable to get in — this alone justifies the tool.
Draft agendas, timelines and division meeting notes
Administrative writing with no student information in it at all.
Explain a pedagogical idea three different ways
Excellent for your own learning — then verify against Ontario sources.
Feature
10 AI tasks that save Ontario teachers the most time
Ranked by the honest gap between how long the task takes by hand and how long it takes to produce and properly review a draft.
Reference
Which tool for which classroom purpose
Every tool below is built for the Ontario context — curriculum terminology, Canadian English and Growing Success language — and every output is a draft for you to review.
Three-part lessons, unit plans and long-range plans mapped to expectations
Diagnostic, formative and summative tasks aligned to the Achievement Chart
September diagnostics that tell you where to start
Levels 1–4 descriptors in language students and families understand
Strength, criterion and next step from evidence you supply
Student-friendly criteria that describe quality
Multiple entry points for the class you actually have
Accommodations and measurable goals from generic descriptions of need
Professional messages home in Ontario teacher voice
Comments that express your judgement in correct terminology
E / G / S / N comments for all six Learning Skills
A full day of coverage written before 7 a.m.
Feature
Questions every teacher should ask before using AI
Eight questions, about ninety seconds. If you can answer all eight, you are on defensible professional ground.
1.Does this protect student privacy?
Reread the prompt as though it will be printed. Could a colleague identify the child? If yes, generalise it.
2.Am I applying professional judgement?
Did I decide first and ask second — or did the tool's suggestion change what I believed about a student?
3.Is this accurate?
Every curriculum reference, policy claim, statistic and citation gets checked against the source.
4.Does this actually support learning?
Or does it just produce more paper? A worksheet nobody needed is not a time saving.
5.Could I explain this decision to a parent?
The transparency test. If explaining the process would be uncomfortable, change the process.
6.Does this align with Growing Success?
Especially for anything touching evaluation, Learning Skills or reporting.
7.Is this appropriate for the students in front of me?
Reading level, cultural relevance, prior knowledge, the six students who need the vocabulary first.
8.Have I read every word of it?
If the honest answer is no, it does not go to students, families or the report card.
What goes wrong
Eight common mistakes and how to fix them
None of these come from teachers who are careless. They come from teachers who are tired in June, which is exactly why the fixes are rules rather than intentions.
1.Entering identifiable student information
Why it costs you: The single most serious risk in classroom AI use, and the easiest to avoid.
The fix: Describe the need, not the child. 'A Grade 5 student who…' is always enough context.
2.Accepting output without review
Why it costs you: AI is fluent and confidently wrong. Errors sound exactly like everything else.
The fix: Rule, not willpower: nothing unread reaches a student, a family or a report card.
3.Using AI instead of professional judgement
Why it costs you: Inverts the relationship and produces decisions you cannot defend.
The fix: Decide first, then ask AI to help you word the decision you already made.
4.Ignoring bias in outputs
Why it costs you: Models reproduce their training data, including whose stories are told and how.
The fix: Read examples, names and contexts through an equity lens before they reach students.
5.Not checking curriculum alignment
Why it costs you: Output often assesses something adjacent to the expectation you chose.
The fix: Hold the draft next to the expectation. If they do not match, rewrite the task.
6.Treating AI as an answer rather than a tool
Why it costs you: Produces polished work that does not fit your class, your pacing or your students.
The fix: Treat every output as a first draft from a keen colleague who has never met your class.
7.Using AI for everything
Why it costs you: Some tasks are faster by hand, and some are the thinking itself.
The fix: If fixing the output takes longer than writing it, stop using AI for that task.
8.Failing to teach students responsible use
Why it costs you: They are already using it; silence just removes you from the conversation.
The fix: One lesson on 'confidently wrong', one class agreement, one AI-use line on every task sheet.
AI is not replacing great teaching. It is making room for it.
Artificial Intelligence cannot replace the encouragement a teacher gives on a hard morning, the trust built over a year of small consistencies, or the insight that comes from having watched a child work through something difficult. It does not know your class. It has never read the Ontario Curriculum with your Grade 4s in mind. It cannot decide what a student needs next, and it should never be asked to.
What it can do is give teachers back something they genuinely need: time. Time for the four-minute conference. Time for feedback that says something. Time to notice the student who has gone quiet. The future of Ontario teaching belongs to educators who combine professional expertise, strong relationships, sound judgement and the thoughtful use of technology — in that order.
Ontario Curriculum AI was built for exactly that combination. It helps you plan lessons and units, design assessments and rubrics, differentiate for the class you actually have, draft student feedback, build IEP language, write to families and prepare report cards — in Ontario terminology, aligned to Growing Success, with student privacy protected and your professional judgement at the centre of every decision. That last part is not a limitation of the tool. It is the point of it.
0% read · about 30 min left
Keep going
Apply this to my classroom™
Use AI responsibly on one real task
Apply the framework to a task you are doing this week, with your professional judgment in front.
- Time
- About 15 minutes
- Steps
- 4
- You will finish with
- One AI-assisted piece of work you have reviewed, edited and would sign your name to.
Opens a guided workshop. Your progress and reflections stay private on this device.
Adapt This for My Classroom™
Rewrite this article for the class you actually teach
Journal will take what you have just read and turn it into a plan sized for your grades, your subjects and the students in your room — with the differentiation moves already worked out.
Tell Journal about your classroom once and every article can adapt to it.
Your profile stays on your device. Only the general shape of your class is sent to generate the plan — never a student name. Everything here is a draft for your professional judgment.
Keep going with Ontario Curriculum AI
Everything below is part of the same Ontario knowledge base — guides, tools and templates that build on this article.
Cornerstone guides
- The Complete Ontario Curriculum GuideHow to read, unpack and sequence Ontario expectations from Kindergarten to Grade 12 — strands, codes, revised documents and coverage.
- The Complete Growing Success GuideOntario's assessment, evaluation and reporting policy in full — levels, qualifiers, learning skills and the rules that govern a mark.
Resources and support
- Download the matching Ontario templateEditable planning and reporting templates built for Ontario classrooms.
- Free Ontario teacher resources and templatesComment banks, rubrics, planning organizers and parent communication templates.
- Help Centre guides for every generatorStep-by-step instructions for each tool, export format and account setting.
- How Ontario Curriculum AI protects student privacyCanadian hosting, PIPEDA alignment and why no identifiable student data is stored.
- Reviewed by Ontario educators
- Updated on a 90-day review cycle
- Built in Ontario
- Growing Success aligned
- Privacy-first — no student data stored
- Evidence-informed
Reviewed each term as Ontario AI guidance develops
Ontario Curriculum AI Newsletter
Get New Ontario AI, Assessment and Reporting Resources First
- Exclusive editable templates
- New AI tools as they ship
- Ontario teacher guides
- Free downloads
- Seasonal resources for each reporting period
Get Ontario Reporting & Planning Updates
New tools, reporting-season guides, and policy insights for Ontario educators.
No spam. Ontario-focused updates only. Unsubscribe anytime.
Ready to get these hours back?
Everything in this article can be drafted for your grade, subject and class in a single sitting. You still apply your professional judgment — the tool just removes the blank page.
- Built for Ontario, not adapted to it
- Growing Success aligned language
- No student data required
- Canadian hosting, PIPEDA aligned
- Free plan available
Teachers also read
- The Ultimate Ontario Lesson Planning WorkflowThe planning system AI-assisted drafting should fit inside.
- Preparing Report Card Evidence Before School StartsGather evidence first; AI only helps you word it.
- Parent Communication Templates Every Ontario Teacher NeedsWording for the family conversation about AI and privacy.
- Learning Goals vs Success CriteriaWhat AI-drafted criteria need to get right before students see them.
- Ontario Assessment StrategiesTriangulated evidence — the strongest academic integrity architecture there is.
- Growing Success GuideProfessional judgement, evaluation and reporting in Ontario policy.
- The Ultimate Ontario Teacher Back-to-School ChecklistWhere to set your AI expectations before the first week begins.
- Ontario Education UpdatesAI, curriculum and assessment policy changes as they are published.
Recommended learning journey
You just read about You now have a defensible personal standard for using AI in an Ontario classroom.. Here is what comes next.
Editorial information
- Reviewed
- August 23, 2026
- Last updated
- August 23, 2026
- Reading time
- 30 minutes
- Difficulty
- practical
- Ontario grade levels
- Kindergarten to Grade 8
- Applicable subjects
- All subjects
- Applicable AI tools
- Lesson and Unit PlannerAssessment GeneratorStudent Feedback GeneratorReport Card Comment GeneratorDifferentiation ToolsIEP Builder
- Evidence level
- Ontario Ministry policy (Growing Success)
- Editorial confidence
- High — verified against current Ministry documents
Every draft produced by Ontario Curriculum AI is a starting point. Ontario educators remain responsible for professional judgment, board policy and the accuracy of anything sent home.