AI Hallucination: A Survival Guide for People Who Publish Under Their Own Name

I think hallucination is the most dangerous thing about AI for anyone who publishes, because it looks exactly like competence. The prose is clean, the citation’s formatted correctly, and the fact is fiction. If your name goes on the work, the checking is your job, and no one else is going to do it for you.

This guide is for anyone publishing under their own name who’s using AI as part of the process. Writers, consultants, researchers, executives writing books or articles, attorneys drafting briefs, anyone whose reputation rides on whether the facts in their work are real. The questions are the ones I get asked most. The answers are short on purpose.

What is an AI hallucination?

A hallucination is any output where an AI states something as fact that isn’t true, with full confidence and no signal to the user that the system isn’t sure.

Hallucinations are invented citations, fabricated quotes, wrong dates, nonexistent studies, made-up legal cases, plausible-sounding statistics with no source. The output reads completely normal. The fact it states is just not real. For more, see AI isn’t coming for your job. Other people using AI are.

Watch: COURT: Utah attorneys sanctioned after Chat GPT fabricates fake case law, KSL News Utah

Why does AI hallucinate?

Large language models work by predicting the next likely word given everything that came before.

They’re not databases. They have no concept of “I know this” versus “I don’t know this.” When a model is asked for a fact it doesn’t have, it generates the most plausible-sounding word, then the next, then the next. The result is a fluent sentence that may or may not be true, and the model has no internal signal that distinguishes the two. For more, see how to turn your AI into the mother-in-Law from hell.

Nothing’s malfunctioning when this happens. The model was built to produce text that sounds right, and it does that very well. Whether the text is right is a separate question it was never built to answer, and I think the companies selling these tools ought to say so a lot louder than they do.

How often do AI hallucinations happen?

Often. Multiple independent studies have measured it. Published court cases have documented citation invention in legal briefs. Statistical invention happens in nearly every long document where you ask the model for specific numbers you didn’t give it. The rate varies by model, by domain, and by how confidently you push the model into territory it doesn’t know, but the baseline is high enough that you should assume any specific claim in AI output may be invented until you verify it.

What kinds of facts does AI invent most?

The pattern is consistent across models. The most common categories:

• ► Academic citations. Authors, journals, page numbers, study titles, all invented with the right shape and zero existence.
• ► Quotes attributed to real people. The person exists. The quote doesn’t.
• ► Statistics with specific decimal places. The more precise the number sounds, the more confident the invention.
• ► Dates of events. Off by months, years, sometimes decades.
• ► Legal cases. Plaintiff, defendant, court, year, the entire citation fabricated.
• ► Book titles and authors that don’t exist.
• ► Technical specifications, especially for older or less common products.

How do I spot a hallucination in my own draft?

Read every specific claim and ask one question: where did this come from?

If the answer is “the AI said it” and there’s no source the AI gave you that you can independently verify, treat the claim as unverified. That means every name, every date, every number, every quote, every citation. Especially the ones that sound the most authoritative, because the model produces its most confident invention exactly when it’s the least to work from. If you run anything through ChatGPT or any other AI, fact-check it to death. It’ll put you at the wrong company or in the wrong place, change your name, change your links, and sometimes the errors are so subtle you’ll read right past them.

Specific tells worth scanning for:

• ► Citations that name a researcher and a paper but no DOI or URL.
• ► Statistics rounded to weird decimal places (43.7%, 28.3%) that you can’t trace to a source.
• ► Quotes that sound exactly like what the model would have written, attributed to someone famous.
• ► Dates that don’t match what you can find elsewhere.
• ► Any reference to a “recent study” with no specific journal or year.

Will turning on web search fix it?

Partially. When the model has live search access, it can pull actual sources and cite them, and the citation invention rate drops significantly. The problem is that the model still synthesizes the result, and the synthesis can still misrepresent what the sources said. So you trade citation invention for paraphrasing errors. They’re harder to catch because the sources are real but what the model says they say isn’t quite accurate.

Web search helps, and I’d turn it on. The verification step is still yours, and it’s where people get lazy, because a real link next to a claim feels like proof. It only proves the page exists. Open it and read what it says.

Can I just ask the AI to fact-check itself?

No. Asking the model “are you sure?” or “is this true?” doesn’t pull from a different knowledge base. It just generates a new response using the same mechanism. Often the model will confidently confirm an invention because the invention sounds plausible to it on a second read.

Some models will flag low confidence on certain outputs, and those flags are worth paying attention to. The absence of a flag isn’t a confirmation of accuracy.

A separate session does work for me. When Claude writes an article, I paste it into a different project I keep just for fact checks and ask for one there. It goes out to the web and tells me what’s wrong and what’s right. It has to be an unconnected project, because you don’t want the context from the first one.

How do you prevent AI hallucinations in your work?

Verify every specific claim against a source you can open yourself, and never publish a quote, number or citation the AI supplied without checking it.

From the podcast

Ben “The Automator” Christensen applies the same standard to AI output that he had to a contractor’s, and says so plainly on Leaders and Their Stories: If you’re a business owner out there and you’ve got this fully automated system, do some QA checks, do some different things along those lines. He QAs the calls on every voice assistant he builds, for exactly the reason this article describes.

Verify every specific claim before publication. There’s no shortcut. The workflow that works:

• ► Generate the draft with AI if that’s part of your process, knowing the draft is a starting point not an authority.
• ► Highlight every factual claim: names, dates, numbers, quotes, citations.
• ► For each one, find an independent source. Not the AI. The original.
• ► If you can’t verify a claim, cut it or replace it with something you can verify.
• ► Ship the verified version.

This is slower than publishing the AI draft, and it’s the only thing that keeps a hallucination off your byline. People bring me books they wrote with AI that turned out to be crap, and they want them fixed. It’s a big market, and I wish it were smaller, because a lot of that pain would never happen if people did this checking before they hit publish.

What if I publish a hallucination by accident?

Fix it publicly and quickly. The standard for handling published mistakes is the same as it was before AI existed: name the error, correct it, explain how it happened. The reputation damage from a corrected mistake is small. The reputation damage from a hallucination that stays uncorrected on your byline is permanent.

Attorneys have been sanctioned for citing AI-invented cases in court filings. Authors have had books withdrawn from publication for fabricated citations the AI introduced. Journalists have lost careers over invented quotes. The cost of catching it before publication is always less than the cost of catching it after.

The professional who blames the software after the fact is passing the buck. You chose to publish it. The model can’t be sanctioned, fired or made to withdraw a book, so every bit of the cost lands on the person whose name is on the work.

AI hallucinates and drifts, and it has assumptions built in, so everything it hands you needs review. I compare it to the AI-drafted visit notes clinics use now, the chart a doctor makes after seeing you. The AI makes mistakes there too. If it charts somebody with issues they don’t have and nobody reviews it, somebody could get hurt.

Checklist before you publish anything AI helped you write

• ► Every citation traced to an independent source you clicked on.
• ► Every quote checked against the speaker’s actual published or recorded statements.
• ► Every statistic verified against the original study, not a secondary mention.
• ► Every date confirmed against a primary source.
• ► Every name spelled the way it’s spelled.
• ► Every legal case verified in a legal database, not just confirmed by a second AI prompt.

If you can’t check the box, cut the claim. Publishing an unchecked AI fact under your own name is careless, and readers are right to hold it against the author, because one fake quote makes them doubt every real one.

I’d cut a good paragraph before I’d risk a single invented fact in a client’s book. Use the machine as the tool it is and check everything it hands you.

If you want the broader case for why your work has to clear this bar, the piece on why AI alone isn’t the problem walks through what readers catch and how they catch it.

From the author

The Death of Thinking
Verification is a thinking skill, and skills fade when nobody uses them. The Death of Thinking is my book on what outsourcing judgment costs the people who publish under their own names.

Frequently Asked Questions