An online course landing page is built to convert, and the claims on it — “accredited,” “94% of graduates get hired,” “taught by a former [well-known company] engineer,” “money-back guarantee” — are marketing copy first. Before paying, the natural instinct now is to ask an AI assistant “is this course legitimate?” and treat the answer as a check. That instinct has a specific problem: the model has no reliable way to know the current accreditation status, refund terms, or hiring statistics of a specific course, and it may answer fluently and wrongly rather than say “I don’t know.”
This article is a verification checklist for exactly the claims that matter before you pay, plus where AI is actually useful in the process — organizing the checklist, not answering it from memory.
The concept: verification, not recall
A model’s training data has a cutoff and is not a live registry of every course provider’s current accreditation, staffing, or policies. Asking “is [specific course] accredited?” invites the model to produce a plausible-sounding answer based on general patterns — what accreditation claims for similar courses usually look like — rather than a checked fact about this specific course today. The fix is not avoiding AI. It is asking it to do the part it is actually good at: turning a landing page’s claims into a specific, checkable list, and pointing you to the kind of primary source that would confirm or refute each one.
The common misconception
The mistake is treating “I asked an AI chatbot and it said the course seems legitimate” as equivalent to verification. It is not. A model asked a vague legitimacy question will often produce a reasonable-sounding, hedged answer that feels like due diligence but checked nothing — because it cannot browse the accrediting body’s actual current registry, cannot see the course’s actual current refund terms, and does not have a live record of the specific instructor’s current employment.
Never treat an AI tool’s unverified statement about a course’s accreditation as fact. Check accreditation directly against the accrediting body’s own public registry or directory — most legitimate accrediting bodies publish one. A course claiming accreditation from a body that does not list it, or from a body that does not actually accredit that type of program, is a serious red flag regardless of how confidently anything, human or AI, describes it as legitimate.
The four claims worth verifying, and where to check each one
1. Accreditation. Find the specific accrediting body named on the course page. Go to that body’s own website and search its public list of accredited institutions or programs directly — do not rely on the course’s own claim or a search-engine summary of it. There are also registries that let you check the accreditor itself: in the US, the Department of Education publishes the Database of Accredited Postsecondary Institutions and Programs, which lists recognized accrediting agencies alongside the institutions and programs they accredit. Elsewhere, the equivalent is usually your national ministry of education or the relevant professional regulator. If the course names no specific accrediting body, “accredited” without a named body is not a verifiable claim at all.
2. Refund policy. Read the actual terms-of-service or refund-policy page, not the marketing headline. “Money-back guarantee” pages often carry conditions — a strict day window, a requirement to complete a minimum percentage of the course first, or exclusions for certain course types. Screenshot or save the policy as it exists on the day you enroll, since these pages can change.
3. Instructor credentials. Verify the instructor’s stated background independently — a LinkedIn profile, a company’s own team page, a university faculty directory, or a conference speaker bio — rather than trusting only the course page’s own biography of them, which the course provider wrote and has an incentive to present favorably.
4. Outcome statistics. A hiring-rate or income claim (“94% employed within 6 months”) needs a methodology to mean anything: who was surveyed, what counts as “employed” or “in the field,” and what the response rate was. Ask for it in writing. In the US, an advertiser is expected to hold a reasonable basis for an objective claim before making it, not to assemble support afterward if challenged — the principle the Federal Trade Commission set out in its policy statement on advertising substantiation and applies through its general advertising and marketing guidance. That does not tell you whether a specific provider is complying, and rules vary by country; what it tells you is that a provider unwilling to describe the methodology behind a number it chose to advertise is declining to show something it should already have.
Use AI to build the checklist, not to answer it
Here is the marketing copy from an online course landing page: [paste the page text].
Extract every specific, checkable claim about accreditation, refund terms, instructor credentials, and outcome statistics.
For each claim, tell me what primary source I would need to check to verify it (e.g., "the named accrediting body's public registry," "the course's own terms-of-service page," "the instructor's LinkedIn or employer's team page").
Do not tell me whether the claims are true — I will check each source myself.
This produces a specific, organized checklist from the actual page in front of you, without asking the model to guess at facts it cannot reliably know.
An illustrative claim check
The following is an illustrative scenario with invented specifics, not a real course. A page states: “Accredited by the National Council for X. 92% of graduates are hired within 3 months. Taught by a former senior engineer at [well-known company].” Running the extraction prompt above produces three checkable items. Checking the named council’s own public registry finds the specific program is not listed, though a different program from the same provider is — worth a direct question to the provider before enrolling. The refund page, read in full, requires a request within a short window and before starting more than a set fraction of the material: a real guarantee, but a narrower one than the marketing headline implied. The instructor’s profile confirms the stated role, held several years before the current listing, which is accurate but worth noting if currency matters to your decision.
None of that verification required trusting an AI tool’s opinion of the course. It required AI organizing the claims, and primary sources answering them.
Courses tied to a regulated profession need a stricter check
If the course claims to prepare you for a licensed or regulated profession — nursing, real estate, financial advising, teaching — accreditation stops being a quality signal and becomes a gate: it can determine whether the credential is recognized at all by the licensing body in your jurisdiction. For these courses, check directly with your jurisdiction’s licensing board which providers and programs it actually recognizes, before checking anything on the course’s own marketing page. A course can be a genuinely good learning experience and still not satisfy a licensing requirement if it is not on the board’s recognized list — two separate questions that are easy to conflate when a landing page implies otherwise.
Reading reviews without over-trusting an AI-generated summary
If you ask AI to summarize what people say about a course online, the same hallucination risk from why AI sometimes gives confident wrong answers applies to review summaries specifically: a model may blend a small number of reviews it can access into a confident-sounding consensus, miss that most reviews are old or from a different course version, or occasionally invent a plausible-sounding but unverifiable quote. Ask the model to link the specific reviews it is summarizing, and spot-check at least two or three original sources yourself before treating a “reviews are generally positive” summary as settled.
When AI is the right tool for the broader research
If you are comparing several course providers rather than verifying claims on one specific page, Deep Research mode is a better fit for surveying reviews, forum discussions, and comparison articles across many sources at once. This checklist is for the narrower, final step: confirming the specific claims on the specific course you are about to pay for.
Try it today
Pull up a course you are actually considering. Run the extraction prompt on its landing page, then check each item against its actual primary source — the accrediting body’s registry, the real refund terms, the instructor’s independent profile. The online course claim-check worksheet turns this into a one-page checklist you can reuse for the next course you consider.
