Affiliate + AI money courses: the passive-commission claim

You are deciding whether “no-inventory affiliate funnel + AI money course → passive commissions” is durable income or compliance theater with a refund cliff. The scene is familiar: ClickFunnels-style squeeze pages, AI-written VSLs and email sequences, Stripe or ClickBank screenshots, “AI makes money while you sleep” course funnels. The pitch bundles three promises — no inventory because someone else’s product pays the commission, generative tools write the sales assets, and traffic compounds into passive income. Course sellers treat one launch month or a redacted dashboard as proof.

Evidence grade beats vibe. Counts: multi-month net P&L after refunds, chargebacks, ad spend, ESP, and landing tools; disclosed material connections on every endorsement; affiliate-network settlements past cooling-off; ad-account history without serial income-claim disapprovals. Does not count: course testimonial reels, one viral webinar week, redacted Stripe shots, AI-generated “case studies,” or someone else’s affiliate dashboard with the refund column cropped. Launch montages are theater.

Money still follows attention you already own — or can buy at a CAC below settled LTV — plus an offer people keep after cooling-off. Affiliate commissions are downstream of that attention and of settlement (refunds claw commissions back). AI compresses drafting: VSL outlines, email sequences, landing variants, ad angles. It does not invent a buyer list, erase FTC/ASA disclosure duties, make atypical earnings look typical, or turn cold paid traffic into sleep money. Operators with an audience and a disclosed, product-first offer sometimes cut copy time. Operators whose first asset is a “make money with AI” course receipt just ship more funnels that compete with every other AI-written funnel — then share the same refund and ad-policy cliffs.

Regulators and platforms draw a hard line. U.S. FTC Endorsement Guides and Disclosures 101 require clear, conspicuous disclosure of material connections (including affiliate commissions); atypical results need generally expected performance — “results not typical” is not a magic phrase; you cannot invent experience you did not have. ASA/CAP treats affiliate content as advertising that must be obviously identifiable, holds brand and affiliate jointly responsible, and requires business-opportunity earnings claims to be representative — testimonials alone do not substantiate results. Google Ads’ unreliable-claims policy bars get-rich-quick / unrealistic money-making promises even when an exceptional result is technically possible. Meta’s Ad Standards prohibit deceptive or exaggerated success claims used to push commercial schemes. That is the operating environment for “AI makes money” funnels, not optional fine print.

Costs stack quietly. Visible: course and mastermind stacks, ClickFunnels-class landing tools, ESP seats, creative tools (ChatGPT / Jasper / Copy.ai / Writesonic), ad spend. Invisible (and worse): refund and chargeback cliffs after affiliate cooling-off (industry pattern: multi-week hold — ClickFunnels documents ~45 days); ad-account bans when income screenshots and “passive” framing trip misrepresentation rules; compliance rewrites when disclosure was missing; opportunity cost of buying the next course instead of shipping a real offer. Gross commission dashboards look cheap until net after clawbacks and CAC.

Failure modes: course-to-sell-courses loop (the product is hope, not a merchant catalog); undisclosed endorsements, fake scarcity, AI-fabricated case studies, atypical earnings sold as typical; refund cliffs that erase months green on day three; cold-traffic CAC greater than settled LTV against every other AI funnel; serial ad disapprovals and business-manager heat; identical AI VSLs flooding the same verticals.

A useful ledger is boring: traffic source, offer URL, disclosure present (Y/N), ad + tool + course cost, gross affiliate credit, refunds/chargebacks in the cooling-off window, settled net, hours spent. Run it for ninety days across at least one refund cycle. If settled net after your wage rate is negative — or the only “win” was selling another money course — you ran a funnel experiment, not a business. Courses that skip the ledger sell hope; ToolGap treats the ledger as the scoreboard that survives a chargeback.

Who should skip: no audience and no cold-ad budget discipline; income thesis is “passive commissions from AI pages”; plan starts with a stack of AI-money courses instead of a disclosed product-first offer; willingness to use fake scarcity, fabricated case studies, or atypical earnings without expected-results context; first asset is a course receipt and a funnel template. If “passive” is the product promise, walk away — affiliate is settlement risk dressed as sleep money.

Who can pilot small (not the claim as sold): people who already have traffic or an email list and a real merchant or SaaS offer they can disclose. Use AI to cut VSL/email draft time; keep humans on claims substantiation and disclosures; refuse income-screenshot ads that trip Google/Meta unreliable-claim rules; cap the experiment — one offer, fixed ad + tool budget, ninety days, ledger of settled commissions versus refunds. Expand only if settled net beats the prior non-funnel baseline. That is ops leverage on an existing audience, not an AI money-course stack.

Verdict: Skip. Treat “no-inventory AI affiliate funnel → passive commissions” as reliability theater against FTC/ASA endorsement rules, Google/Meta income-claim policies, and multi-week refund settlement. If you cannot earn attention without a course funnel about earning attention, a course funnel will not print settled money for you.