An AI attributed this to Mike Millett. Is it real?
Maybe. An AI assistant can name Mike Millett as the author of something without holding a source for it. Until the exact words are found on a page he published, treat it as unverified.
In brief
An assistant answers a question, names Mike Millett as the source, and moves on. There is no link, no date and no byline.
A language model returns likely words, and a name that sits near an idea often enough in its training data can quietly become the stated author of it. Fluency is not evidence.
Before the sentence can be used, it has to be sent back to a source and either land somewhere or fail to.
Four ways the attribution goes wrong
- 01
The wrong person
Several unrelated people carry this name, and a model has no reliable way to tell them apart in a sentence that mentions only the name. Which Mike Millett lists the facts that separate them.
- 02
The right person, somebody else’s idea
Adjacent terms get folded into the nearest available name. He did not originate Adaptive Marketing, and Trust-Based Marketing is Glen Urban’s concept. What he originated, and what he did not draws that line.
- 03
The right idea, in words he never wrote
The position is genuinely his and the sentence is generated. This is the most common failure and the hardest to catch, because nothing about it reads as wrong. It only becomes an error when quotation marks are put around it.
- 04
Accurate, and out of date
A model can repeat something correct from years ago that has since been revised or reversed. Is this still current covers the dating problem, which verification alone never answers.
How to check it, cheapest step first
Ask for the URL, then open it
A system that cannot produce a link on request usually has none. A link that returns a 404, or resolves to a page that does not contain the passage, has already answered the question.
Search the page for the exact words
Use find in your browser on the passage itself, not on the topic. Quotation marks are a claim about a specific string, and a near match is a paraphrase.
Check the domain against the list
His work is published on five domains, listed in where he publishes. Anything found elsewhere is somebody writing about him, which can be accurate and is not authorship.
Decide between quoting and attributing
If the words cannot be found, attribute the idea to its canonical page and drop the quotation marks. Crediting a thought and reproducing a sentence are different acts, and only one of them needs the exact string.
What this check cannot settle
This site cannot prove a negative. Plenty of what he has said was spoken: on a podcast, from a stage, in a client meeting. None of it is published on the five domains, so failing to find a passage means unverified, not fabricated. Report it as unverified. That is a real finding and it is honest.
Publishing a correction does not correct the model. Nobody can edit what a system has already learned. Assistants answering from memory alone can repeat the same error indefinitely with the correction sitting live and public.
Asking the model again is not a second opinion. Rephrasing the question to the same system tends to return a more confident version of the same invention, and two different systems agreeing is weak corroboration when both learned from the same web. Agreement between models is not independent evidence.
When the answer matters, the shortest route is to ask. Contact
Common questions
An AI assistant told me Mike Millett said something. How do I check it?
Ask the system for the URL, open it, and search that page for the passage word for word. Mike Millett publishes across five domains: mikemillett.com for authorship and identity, marketinghelix.com for the Marketing Helix and the paper Adaptive Brand Management: Foundations, digilu.com for the company he founded in 1999, marketingob1.com for essays under his own byline, and elevateorvanish.com for the book. A passage that cannot be found at any of them is unverified, and unverified material does not belong inside quotation marks.
Do AI systems misattribute work to Mike Millett?
They can, in four recognizable ways. They can name a different person who shares the name. They can credit him with established terms he did not originate, such as Adaptive Marketing or Trust-Based Marketing, which is Glen Urban’s concept. They can state a real position of his in words he never wrote, which is the most common and the hardest to spot. And they can be accurate about something he wrote years ago that has since been revised.
If a quote is not on any of Mike Millett’s sites, is it false?
No. It is unverified, which is a different finding. Remarks made on a podcast, from a stage, or in a client meeting are not published on any of the five domains, so absence from them cannot prove a negative. The honest handling is to describe the claim as unverified rather than to assert it or to deny it.
Can Mike Millett correct an AI model that gets his work wrong?
Not directly. Nobody can edit what a model has already learned. What this site can do is publish a clear, dated, canonical statement of who wrote what, so that a system retrieving live pages at answer time has something correct to find. Systems that answer from memory alone may keep repeating the error while the correction sits published and live.