Launch App
Vetting AI Tools for Schools: Alberta's Playbook, Explained — and Answered
Every AI company wants to be in classrooms. Very few publish their answers to the questions school boards are actually told to ask. As an Alberta company building an AI tutor, we read the ASBA guidance the way a restaurant should read the health code — not as an obstacle, but as the standard you should be proud to be measured against. Here's what the guidance says, why it's good policy, and our own scorecard against it.
What ASBA's AI Policy Guidance is
The ASBA AI Policy Guidance is a document from the Alberta School Boards Association that helps Alberta's elected school boards write their own policies for artificial intelligence in K-12 education. Alberta's boards set their own policies — so rather than dictating rules, ASBA equips them with principles, practical guidelines, and a review checklist for their existing policies.
It's a serious piece of work. The principles come from University of Alberta research on ethical AI in K-12 education (which itself builds on UNESCO, UNICEF, and OECD ethics work), and the practical guidance draws on TeachAI's guidance toolkit. The result is human-centric without being anti-technology: it takes AI's benefits for learning seriously and treats the risks as things to manage deliberately, not wish away.
ASBA has also backed this up structurally — it created an Innovation and Artificial Intelligence Committee of directors and trustees to keep guiding members as the technology evolves. For anyone building AI for classrooms, that's the audience whose questions you should be able to answer.
What it asks school boards to do
In one paragraph: the guidance asks boards to build AI literacy for staff and students, engage their communities, update academic-integrity policies for the AI era, ensure equitable access so AI doesn't widen the digital divide, and put real structure behind all of it — a responsible-AI committee, an inventory of AI tools in use, humans in the loop, cybersecurity that accounts for AI, and data-privacy compliance under Alberta's public-sector privacy law (FOIP).
Two ideas stand out. First, the guidance treats AI as a socio-technical system — meaning you can't evaluate the technology apart from the people and context around it. A tutoring tool isn't just a model; it's a model plus a child, a family, a teacher, and a set of incentives. We think that's the single most clear-eyed sentence-level idea in the document. Second, it insists on age-appropriateness, noting that most free generative AI systems were never designed for children — ChatGPT's own terms of use don't allow users under 13, and its content isn't always appropriate for minors.
The vetting questions — and why they're the right ones
The heart of the guidance for any tool-builder is the vetting process. ASBA recommends that boards assess every AI tool against established criteria, keep the results transparent to stakeholders, and maintain a pre-approved "safe list" of trusted tools so teachers can pick technology that's already been reviewed. Here's what that assessment covers, in plain language:
| Vetting area | What a board wants to know |
|---|---|
| Data privacy | What data is collected, where it goes, who can see it, and whether handling complies with privacy law. |
| Age-appropriateness | Was this designed for children? What are the age rules, and how is parental consent handled? |
| Academic integrity | Does the tool support learning, or does it just produce answers a student can copy? |
| Equity & accessibility | Can every student use it — regardless of income, language, or accessibility needs? |
| Algorithmic bias | Has the vendor considered and tested how the system might treat different students differently? |
| Cybersecurity | Is the system hardened — including against newer, AI-specific attacks? |
| Transparency | Will the vendor show its work — and can assessment results be shared with parents and staff? |
If you're a parent, notice something: these are exactly the questions you'd want to ask about any app your child uses — the board process just formalizes them. And if you're an AI vendor, the implication is blunt: if you can't answer these questions in writing, you shouldn't be in a classroom.
Our answers: MapleMind against the checklist
MapleMind: AI Homework Tutor is a Canadian AI tutoring app for K-12 students (Grades 1–12, all 10 provinces) covering Math, Science, English, French, and Social Studies — with step-by-step homework help, adaptive practice, and provincial exam prep (EQAO, FSA, PAT). We're an Alberta company, so this guidance describes our home turf. Here is our own product, measured against ASBA's list:
| Vetting area | MapleMind's answer |
|---|---|
| Data privacy | Built in Canada and PIPEDA-compliant. Student AI data is protected by a data-processing agreement with our AI provider and isn't used to train their models. No ads. No sale of personal information. Parents can export or delete their child's data. |
| Age-appropriateness | Designed for K-12 from day one: an age gate, and verified parental consent for users under 13 — the parent, not the child, holds the account controls. This is the direct opposite of general chatbots whose terms simply exclude young users. |
| Academic integrity | Guided Mode teaches the concept and works a different example instead of answering the assigned question — and switching it off is locked behind the account password or Face ID, so a student can't quietly turn it into an answer machine. It's an integrity policy a device can't un-decide. |
| Equity & accessibility | A free daily tier (5 tutoring chats + 1 practice quiz, resetting every 24 hours), runs on any phone or in any browser at app.trymaplemind.ca (nothing to install on school devices), and works in 14 languages — the interface and the tutor's explanations — including many of the languages Alberta's newcomer families speak at home. |
| Algorithmic bias | Mitigated by design — the tutor is constrained to curriculum-aligned content, shows its reasoning step by step, and makes no automated decisions about students (no grading, ranking, or placement). Formal bias evaluation is on our roadmap, and we won't claim it before it's done. |
| Cybersecurity | A full security audit in July 2026 (endpoint hardening, dependency vulnerabilities cleared, rate limiting, abuse protections), with prompt-hardening against misuse. Independent penetration testing is a planned next step — again, named, not claimed. |
| Transparency | This table is the down payment: we're publishing our vetting answers rather than waiting to be asked. Our privacy policy and terms are public, and we'll gladly walk any board, committee, or parent through the details. |
What we're still working on
A vetting culture only works if vendors are honest about the unfinished parts, so here are ours. We haven't yet completed a formal accessibility (WCAG) audit — the app supports font scaling and 14 languages, but a proper audit is different from good intentions, and it's on the list. Our bias-evaluation approach is currently design-level (curriculum constraint, visible reasoning, no automated decisions) rather than a published testing framework; formalizing it is planned. And for school-board deployments specifically, boards operate under Alberta's FOIP public-sector privacy law — a different regime from the consumer privacy law we're built on — so a board partnership would involve preparing privacy-impact documentation together with that board.
We publish this list on purpose. The guidance asks boards to make assessment results transparent; the least a vendor can do is start by assessing itself in public.
What parents can take from this
You don't need a policy committee to use ASBA's thinking at your own kitchen table. Before your child uses any AI tool, four questions cover most of the ground:
- Who is it for? Check the age rules. If a tool's own terms say 13+, that's the vendor telling you it wasn't built for your ten-year-old.
- What happens to my child's data? Look for plain answers about collection, sharing, and deletion — and be wary if you can't find any.
- Does it teach, or just answer? A tool that hands over finished answers is a shortcut machine. A tool that explains, checks understanding, and works examples is a tutor. (We've written more about that line in our guide to AI homework and cheating.)
- What does the company admit it hasn't done yet? Every honest vendor has an "in progress" list. If a company claims everything is perfect, that's not confidence — that's marketing.
Alberta's boards are asking the right questions. The more parents ask the same ones, the better every tool in this space will have to become — ours included. That's the point of a standard, and it's why we're glad this one exists. (For the wider policy picture, see our plain-language guide to Canada's national AI strategy.)
Frequently asked questions
What is ASBA's AI Policy Guidance?
It's a policy document from the Alberta School Boards Association that helps Alberta's school boards write their own rules for AI in K-12 education. It covers AI literacy, academic integrity, equitable access, data privacy, cybersecurity, copyright, and — critically — how boards should vet AI tools before students use them. It draws on University of Alberta research, TeachAI, the OECD, and UNESCO-linked ethics work.
How do school boards vet an AI tool?
ASBA recommends boards assess each AI tool against established criteria — data privacy, algorithmic bias, cybersecurity, age-appropriateness, accessibility, cost, and general safety — keep an inventory of AI tools in use, and maintain a pre-approved "safe list" so teachers can quickly choose technology that has already been reviewed.
Why aren't free AI chatbots appropriate for young students?
Most free generative AI systems were not designed for children. ASBA's guidance notes, for example, that ChatGPT's own terms of use don't allow users under 13, and require parental permission under 18 — and that its content is not always appropriate for minors. Tools built for kids need age gates, verified parental consent, and content designed for school use.
Does MapleMind meet ASBA's vetting criteria?
MapleMind is built in Canada, PIPEDA-compliant, requires verified parental consent for users under 13, and its Guided Mode — which teaches the concept instead of answering the assigned question — is locked behind a parent's password. Some items are still in progress, like a formal accessibility audit and a published bias-evaluation approach, and we say so openly rather than claim otherwise.
What should parents take from ASBA's AI guidance?
The vetting questions boards ask are the same ones parents can ask about any AI tool their child uses at home: who can use it (age rules), what happens to my child's data, does it teach or just answer, and what does the company admit it hasn't done yet. A trustworthy tool should answer all four plainly.
Sources
- Alberta School Boards Association — AI Policy Guidance v1.0 (2024)
- ASBA — Announcement: AI Policy Guidance for K-12 Education
- Adams et al. (2023) — Ethical principles for artificial intelligence in K-12 education, Computers & Education: Artificial Intelligence
- TeachAI — AI Guidance for Schools Toolkit
- OECD — AI Principles and definition of an AI system
An AI tutor built to pass the vetting questions
MapleMind is a guided, curriculum-aligned AI tutor for every Canadian K-12 student — verified parental consent for under-13s, a locked Guided Mode that teaches instead of answering, and no ads or data sales, ever. Free to start.