FTC Operation AI Comply: what evidence questions should buyers ask?
Last reviewed June 1, 2026
Operation AI Comply is useful to buyers because it groups named FTC AI claim cases in one sweep: AI as a professional substitute, AI-generated reviews, AI-powered ecommerce income, and later AI detector accuracy enforcement. This page stays on that sweep and turns those source-backed patterns into evidence requests a buyer can use before relying on similar vendor wording.
Evidence buyers verify
- The exact public claim wording and dated source where it appeared.
- Records that existed when the claim was published, not only later explanations.
- Task, benchmark, customer cohort, or workflow scope that matches the words in the claim.
Opens the checker for this claim type. Paste your vendor's exact wording there. Evidence questions only — not a blacklist or fraud detector. Not sure what a result looks like? See a sample receipt.
Sources this guide draws from
- FTC Operation AI Comply announcement FTC enforcement· September 25, 2024
Official FTC sweep announcement covering DoNotPay, Rytr, Ecommerce Empire Builders, and FBA Machine / Passive Scaling claim patterns.
- FTC Ascend Ecom case page FTC enforcement· Last updated June 23, 2025
Official FTC case page for AI-powered ecommerce income and passive-income claim wording.
- FTC FBA Machine / Passive Scaling case page FTC enforcement· June 3, 2024
Official FTC case page for AI-powered storefront software and online-store income claim wording.
- FTC Content at Scale AI case page FTC enforcement· August 28, 2025
Official FTC case page for Workado / Content at Scale AI detector accuracy claim records.
Public claims with documented evidence gaps
"the world's first robot lawyer"
Automation / Replacement- Source and date
- FTC Operation AI Comply announcement · September 25, 2024
- Evidence signal
- Professional-substitution wording without a task boundary or qualified review standard.
- Evidence gap
- A buyer needs the legal tasks tested, the professional comparison standard, review involvement, known failure cases, and user notice about when qualified advice remains necessary.
- Buyer question
- For the world's first robot lawyer claim, what legal tasks were tested against qualified review and where does the AI assistance stop?
"generate perfectly valid legal documents in no time"
Automation / Replacement- Source and date
- FTC Operation AI Comply announcement · September 25, 2024
- Evidence signal
- Output-quality and speed wording for legal documents without visible document type, jurisdiction, or review limits.
- Evidence gap
- A buyer needs document categories, jurisdictions, test cases, human-review process, error handling, and excluded uses.
- Buyer question
- For the perfectly valid legal documents claim, which document types and jurisdictions were tested, and who reviewed the output quality?
"Testimonial & Review generation"
Compliance / Safety- Source and date
- FTC Operation AI Comply announcement · September 25, 2024
- Evidence signal
- AI-generated review workflow without evidence that generated claims come from real customer experience.
- Evidence gap
- A buyer needs review source, customer experience record, AI-generation disclosure, moderation policy, and controls preventing unsupported product details.
- Buyer question
- For a testimonial-generation AI feature, what evidence shows each generated review is tied to an actual customer experience and disclosed where required?
"Skip the guesswork and start a million-dollar business today"
ROI / Outcome- Source and date
- FTC Operation AI Comply announcement · September 25, 2024
- Evidence signal
- Large outcome claim connected to an AI-powered ecommerce opportunity without customer outcome distribution.
- Evidence gap
- A buyer needs cohort outcomes, total costs, time period, unsuccessful-store rate, refund rate, and the role of AI versus non-AI coaching or marketplace effects.
- Buyer question
- For the million-dollar business claim, what percentage of comparable customers reached that outcome after total costs and over what period?
"quickly earn thousands of dollars a month in passive income"
ROI / Outcome- Source and date
- FTC Ascend Ecom case page · Last updated June 23, 2025
- Evidence signal
- Monthly income wording tied to AI-powered tools without visible median outcome, customer cohort, or cost basis.
- Evidence gap
- A buyer needs customer sample, median and range of outcomes, time period, inventory and service costs, refund rate, and a non-AI baseline.
- Buyer question
- For the thousands of dollars a month claim, what customer outcome distribution supports the result after total costs?
"make money operating online storefronts using AI-powered software"
ROI / Outcome- Source and date
- FTC FBA Machine / Passive Scaling case page · June 3, 2024
- Evidence signal
- Storefront income wording tied to AI-powered software without visible customer outcome, cost, refund, or failure-rate evidence.
- Evidence gap
- A buyer needs net customer results, total investment, inventory and platform costs, unsuccessful customer rate, refund outcomes, and evidence that AI-powered software changed results.
- Buyer question
- For the AI-powered storefront income claim, what share of customers made net income after software, inventory, advertising, fulfillment, and platform costs?
"98 percent accurate"
Accuracy / Performance- Source and date
- FTC Content at Scale AI case page · August 28, 2025
- Evidence signal
- Headline detector accuracy number without benchmark scope, content categories, or false-positive and false-negative rates.
- Evidence gap
- A buyer needs the benchmark corpus, model versions, content types, sample size, threshold, false positive rate, false negative rate, and date of the supporting record.
- Buyer question
- For the 98 percent accurate claim, what benchmark record supports the number and what false-positive rate applies to our content type?
Match each claim pattern to the evidence buyers need
| Claim pattern | Evidence needed | Buyer question |
|---|---|---|
| AI replaces a professional service or expert role | Task scope, qualified review standard, tested outputs, user warnings, failure handling, and escalation path. | Which professional tasks were tested, and what review boundary prevents over-reliance on AI output? |
| AI creates reviews, testimonials, or social proof | Customer-experience source record, AI-generation disclosure, moderation policy, and controls against unsupported details. | How does the vendor prevent AI-generated review text from presenting unsupported customer experience as real social proof? |
| AI-powered business opportunity or income outcome | Customer cohort, median outcome, unsuccessful-user rate, total cost basis, time period, and AI contribution isolation. | What result remains when non-AI coaching, marketplace, inventory, and ad-spend effects are separated? |
| AI accuracy percentage or detector performance | Benchmark corpus, task scope, model versions, threshold, false positive rate, false negative rate, and record date. | Does the benchmark match the content, model outputs, and review decision we would rely on? |
| AI-powered tool claim used to support a broader outcome | Description of the AI component, what it changes in the workflow, baseline without AI, and measured effect of that component. | Which part of the claimed outcome is attributable to the AI component rather than the surrounding service package? |
Evidence to request
- The exact public claim wording and dated source where it appeared.
- Records that existed when the claim was published, not only later explanations.
- Task, benchmark, customer cohort, or workflow scope that matches the words in the claim.
- Failure cases, excluded use cases, and human-review boundaries where the claim could be overread.
- For income or outcome claims, customer distribution and total cost evidence rather than best-case examples.
Questions to put in front of the vendor
- Which Operation AI Comply pattern does this claim resemble: professional replacement, AI-generated social proof, income outcome, or benchmark accuracy?
- What dated source record supports the exact words used in the claim?
- What evidence existed before the claim was published, and can the vendor produce it now?
- Does the evidence cover the same task, customer type, workflow, and environment described in the public copy?
- Which words carry the highest evidence burden: first, perfectly, valid, in no time, million-dollar, passive income, or a percentage?
Wording boundaries to compare against
- Drafts first-pass documents for specified use cases and routes higher-risk situations to qualified review.
- Generates review drafts only from documented customer feedback, with AI contribution disclosed where required.
- Reported customer outcomes for a named cohort over a stated period, with total costs and unsuccessful outcomes included.
- Reported accuracy on a named benchmark covering specified content types, model versions, threshold, and error rates.
Have your vendor's exact claim wording ready?
Check an Operation AI Comply-style AI claim How the evidence method works