What New AI Guidance Means for Course Creators: A Practical Governance Checklist

When new international guidance on artificial intelligence appears, the instinct is often to rewrite the whole AI policy immediately. For most tutors, course creators, and small learning providers, that is the wrong first move. A better response is to review the real learner journey, identify where AI already affects decisions or data, and make a small number of documented, accountable changes.

UNESCO announced on 25 August 2026 that an Education in the Age of AI ministerial statement was expected to be adopted during Digital Learning Week 2026. The event is scheduled for 8–11 September 2026 in Paris. At the time of writing, that means course creators should treat the announcement as a prompt to prepare and review—not as a final set of rules that has already replaced local requirements or provider obligations.

This distinction matters. International statements can help establish direction and shared language, but your day-to-day decisions still need to fit the places where you operate, the ages and needs of your learners, contractual commitments, platform terms, assessment arrangements, and your own safeguarding and privacy responsibilities.

Why a workflow review beats a rushed policy rewrite

A policy can sound responsible while leaving the practical questions unanswered. For example: Can a tutor paste a learner’s draft into an AI tool? Must a learner disclose AI assistance? Who checks automated feedback before it is sent? What happens when an AI-generated response raises a welfare concern or appears inaccurate?

A workflow review starts with actions rather than slogans. It asks what happens before, during, and after a learner encounters AI. This creates an evidence trail for decisions and makes your rules easier for staff, contractors, learners, and parents to follow.

UNESCO’s existing guidance on generative AI recommends a human-centred and age-appropriate approach, including attention to data privacy and to the ethical validation and pedagogical design of uses. Its teacher competency framework likewise places human agency, inclusion, non-discrimination, and linguistic and cultural diversity at the centre of effective AI use in teaching, learning, and assessment.

For a course business, translate those broad principles into operational questions: What is the educational purpose? Who retains judgement? What information enters the tool? How will learners understand the limits of the output? What is the route for challenge, correction, or escalation?

What to verify after the ministerial statement is adopted

Do not revise your policy from summaries, social posts, or assumptions about what the final text will say. Once the statement is available, assign one person to read the official version and compare it with your existing policy and workflow register.

• Status and scope: Is it a statement of shared principles, a recommendation, a framework, or something else?
• Audience: Does it address governments, institutions, educators, technology providers, learners, or several groups?
• New expectations: Which themes are genuinely new, and which reinforce practices you already have?
• Definitions: Note how the document uses terms such as human oversight, transparency, inclusion, safety, data, assessment, and agency.
• Local fit: Identify which points require a check against your local law, safeguarding procedure, contracts, funder requirements, or exam-provider rules.
• Evidence needed: Decide what records would show that your practice matches the principles you choose to adopt.

Keep the review proportionate. A one-person tutoring practice may need a short workflow register and a learner notice. A school or L&D provider may need named owners, procurement checks, staff guidance, training records, and a formal review cycle.

Map AI across the learner journey

Many organisations only map the visible chatbot. Governance needs a wider view. AI may be involved before enrolment, in course production, in learner support, and in quality assurance. Map each use separately because the risks and controls differ.

Include tools that staff use informally. A tutor using a public AI assistant to improve wording for learner feedback may create a different governance question from a tool embedded in your learning platform. The key is not to ban every experiment by default; it is to make the purpose, boundaries, and decision-maker visible.

Set learner-facing rules that people can actually use

Learners need plain-language expectations before an issue arises. Put the essentials where the activity happens: in the course welcome area, assignment brief, live-session guidance, or feedback policy. Avoid a long policy that learners must interpret alone.

• Disclosure: State when learners should acknowledge AI assistance and what level of detail is expected.
• Acceptable use: Explain which activities allow AI support, which restrict it, and why the distinction exists.
• Human review: Make clear that AI output may be incomplete, inaccurate, biased, or unsuitable for the learner’s context, and that learners remain responsible for submitted work where applicable.
• Support and challenge: Tell learners how to ask for a human explanation, report a concerning output, or flag a potential error.
• Boundaries: Explain that learners should not enter confidential, sensitive, or other protected information into unapproved tools.

UNESCO’s student competency framework describes a human-centred mindset and AI ethics as core dimensions, alongside understanding AI techniques and applications. That is a useful reminder that learner guidance should build judgement, not merely list prohibited prompts.

Review data, recordings, prompts, and third-party tools before launch

Before learners use an AI-supported workflow, document what goes in, what comes out, and who can access each stage. Treat prompts, uploaded documents, chat logs, transcripts, generated summaries, and learner identifiers as separate items in your review. A tool can appear low-risk until a staff member uses it with a real learner case.

Ask: Is this information necessary for the educational purpose? Is there a lower-data alternative? Is the tool approved for this use? Who can retrieve outputs? What is the retention approach? How will you respond if the tool produces harmful, inappropriate, or materially wrong content? Where learners are children or otherwise need additional support, involve the appropriate safeguarding lead before implementation.

These questions support responsible design, but they do not substitute for professional or legal advice. Requirements differ by jurisdiction, learner group, contract, and tool.

AI-guidance response worksheet

Use the worksheet below for each AI-supported workflow. Copy it into your course-operations document, assign an owner, and save the completed version alongside your relevant policy, tool review, or course change record. A publisher can also format this table as a downloadable worksheet for staff use.

A practical 30-day implementation plan

1. Days 1–5: Name an AI-governance owner and list every current or proposed AI workflow.
2. Days 6–10: Complete the worksheet for the highest-impact workflows, especially those involving learner work, feedback, recordings, or assessment.
3. Days 11–15: Confirm where human review is required and write concise learner-facing rules.
4. Days 16–20: Review tools, data flows, access, retention, and escalation routes with the relevant internal leads or advisers.
5. Days 21–25: Brief tutors, contractors, and support staff on the approved workflows and the boundaries that apply.
6. Days 26–30: Publish learner communications, save approval records, and schedule the first review meeting.

The goal is not a perfect policy that anticipates every technical change. It is a repeatable way to make decisions, show who made them, and revise them when evidence, tools, or learner needs change. SubSchool can help teams reduce repetitive teaching administration while educators retain authorship and the final educational decision. Explore SubSchool for Business if you are building more consistent course workflows across a team.