AI-Generated Reviews: How Affiliates Can Use LLMs Without Violating FTC Rules

Introduction — Why this matters now

Affiliates increasingly use large language models (LLMs) to scale review content, but U.S. regulators have moved quickly: the Federal Trade Commission (FTC) now prohibits fake or misleading consumer reviews and can seek civil penalties for violations. The rule targeting fake/AI-generated reviews and related deceptive practices took effect in October 2024 and makes clear that reviews must reflect genuine experiences and that certain undisclosed relationships or manufactured testimonials are unlawful.

Alongside the new rule, the FTC’s longstanding Endorsement Guides require clear disclosure when a reviewer has a material connection to the brand (payments, free products, affiliate relationships, employment, etc.). If you use LLMs to assist writing reviews, you must design workflows that preserve authenticity, disclose material connections clearly, and avoid practices the FTC now bans (including buying or fabricating reviews).

Compliance Checklist — Practical do's & don'ts for affiliates

1. Do not publish fabricated or non‑experiential reviews. Any review that misrepresents an actual user experience — including reviews created for a nonexistent person or produced solely by AI without a real experience — is prohibited.
2. Disclose material connections clearly and conspicuously. If you receive affiliate commissions, free product, or other incentives, disclose that relationship so readers can weigh the endorsement. Short-form disclosures must still be prominent and understandable.
3. If you use an LLM to draft or edit a review, label the assistance. Be transparent: state that portions were AI-assisted and confirm the reviewer’s real experience (e.g., “Drafted with AI assistance; I personally used this product for X weeks”).
4. Never condition incentives on positive sentiment. Offers that pay for reviews only if they are positive (explicitly or implicitly) are banned.
5. Avoid insider reviews without disclosure. Company employees, officers, or managers writing reviews must have their relationships disclosed; distributing such content without disclosure can violate the rule.
6. Keep records and provenance logs. Maintain documentation showing product receipt/purchase, who wrote or edited the review, AI prompt and outputs, timestamps, and human sign‑off to demonstrate good‑faith compliance if questioned.
7. Use human verification gates. Require a human who actually used the product to review and confirm any AI‑drafted text before publication. Maintain version history showing human edits.
8. Don’t repurpose unrelated reviews. Reusing a review from one product and presenting it as feedback for another (a practice sometimes called review hijacking) is prohibited.

Quick do/don't table

Disclosure templates & implementation workflow

Below are practical disclosure templates you can adapt. Use the shortest template for tweets/short captions, the medium for video descriptions, and the long form on dedicated review pages. These are examples — tailor them to your facts and keep disclosures prominent (not buried in a link or tiny font).

Short (social post / caption)

Disclosure: I may earn commissions from purchases. This review was edited with AI assistance; I personally used this product.

Medium (video description / blog sidebar)

Disclosure: This post includes affiliate links and I may receive a commission if you buy. Portions of this review were drafted with AI assistance; the reviewer personally tested the product for X weeks and verified the content.

Long (dedicated review page)

Full disclosure: I received this product [free/discounted] from [Brand] or purchased it on [date]. I may earn an affiliate commission for purchases made through links on this page. This review was produced using the following workflow: (1) reviewer personally used the product from [date] to [date]; (2) initial draft was created/edited with an LLM (model name optional); (3) all final text was reviewed and approved by the named reviewer, who confirms the factual accuracy and first‑hand experience. Contact: [email].

Implementation workflow (recommended)

1. Acquisition proof: Record purchase receipts or product shipment/serial numbers where possible.
2. Testing phase: Require a minimum usage period appropriate for the product category (e.g., 2 weeks for a gadget, 30 days for skincare).
3. Drafting: Use LLMs only to generate draft copy, headlines, or alternative phrasings — mark generated outputs in the content system.
4. Human review & attestation: A named human reviewer who used the product must edit and sign off on the review, confirming accuracy.
5. Disclosure insertion: Place visible disclosure near the top of the review, in any video overlay or caption, and in structured metadata (e.g., meta description or schema).
6. Recordkeeping & audits: Log prompts, LLM outputs, human edits, and provenance records for at least several years to respond to enforcement inquiries. The FTC's guidance anticipates documentation may be relevant in investigations.

Enforcement snapshot. The FTC has already used its new rule in enforcement (e.g., orders against platforms and businesses that misrepresented reviews), signaling active oversight — treat compliance as a material operational risk.

Final notes & risk management

LLMs can speed content production, but the legal line is clear: do not present AI‑generated text as genuine consumer testimony if there is no real user experience, and always disclose material relationships. Incorporate these templates into your CMS, train creators on disclosure language, and build simple internal audits to detect problematic patterns (bursts of highly similar five‑star reviews, reliance on purchased review services, or repeated insider postings). When in doubt about a specific practice, consult legal counsel — this guidance is practical and educational, not legal advice.