23/07/2026
The Dangerous Side of AI Nobody Is Talking About
Artificial intelligence has made creating an information product dramatically easier.
Today, you can use AI to help research a subject, organise your ideas, write a first draft and even produce the design.
Within a few hours, you could have an ebook, guide, newsletter or digital course ready to publish, without printing, storing or delivering a single physical product.
This means almost anyone can now create a home-based publishing business.
But there is a major problem…
The barriers to entry have come tumbling down.
And the floodgates have opened.
The internet is becoming overwhelmed with low-quality, AI-
generated material produced by people who appear to care more
about making a quick sale than helping the person who buys it.
It has been given a rather unflattering name:
AI Slop.
Much of it is bland, repetitive and instantly forgettable.
But in certain subjects, it can be far more serious than that.
It can be dangerous.
When AI Slop Becomes Dangerous
In 2025, the AI-detection company Originality.ai examined 558
books published in Amazon’s herbal-remedies category.
It concluded that 82% were likely to have been written using AI.
That does not prove that every book was entirely AI-generated.
Detection tools can make mistakes and should never be treated as
infallible.
Nevertheless, the scale of the findings was alarming, particularly
because these books were offering health-related information.
Liverpool medical herbalist Sue Sprung warned that a vast amount
of unreliable herbal research already exists and that AI may be
unable to distinguish useful evidence from worthless material.
That is the heart of the problem.
An AI system can produce a confident, convincing answer without
genuinely understanding whether the advice is accurate, safe or
appropriate for the person reading it.
That may be irritating if the subject is decluttering your wardrobe.
It is much more worrying if the subject is herbal medicine, mental
health or identifying poisonous mushrooms.
Imagine an AI-generated book recommending a dangerous plant,
an unsafe dosage or a remedy that interacts with someone’s
medication.
The words may look polished.
The cover may look professional.
The author may appear knowledgeable.
But none of that guarantees that the information is safe.
A Book That Should Never Have Been Published
In 2025, Richard Wordsworth went online to find out more after
being diagnosed with adult ADHD.
Amazon recommended a book about the condition.
However, when he began reading it, he discovered obvious
historical inaccuracies and some deeply troubling advice.
Among other things, the book suggested that a person’s friends
and family might never forgive the emotional damage caused by
their ADHD.
It also included a frightening claim about the likelihood of dying
early.
Imagine receiving that message shortly after being diagnosed.
Someone looking for reassurance, understanding and practical
support could instead be left feeling ashamed, frightened or
hopeless.
This wasn’t a poorly written article about an inconsequential
subject.
It was unverified mental-health advice being sold to potentially
vulnerable people.
That is where careless use of AI crosses an important line.
The Publisher Must Take Responsibility
The purpose of a good information product is to save the reader
time and uncertainty.
They buy it because they expect the creator to have:
Researched the subject
Compared different sources
Removed inaccurate or irrelevant information
Organised the useful material
Presented it clearly
Considered whether the advice works in real life
In other words, they are paying you to sift through the rubbish on
their behalf.
AI can help enormously with that work.
But it cannot accept responsibility for the finished product.
You must do that.
So, if you’re thinking about creating an ebook, newsletter, course or
membership using AI, I want to show you two things:
How to avoid publishing inaccurate, misleading or potentially
dangerous information.
How to make your content stand out from the growing tide of
generic AI material.
It starts with one important rule.
Think Like a Publisher
You don’t necessarily need to be the world’s leading expert to
create a useful information product.
You don’t need to be a professional writer.
And you don’t always need formal qualifications.
You can create value by researching, gathering, organising and
clearly presenting useful information to a particular audience.
But the moment you publish that information, you become
responsible for it.
You are the publisher.
Your name, or your business name, is attached to the product.
The buck stops with you.
That means making a reasonable effort to ensure:
The facts are accurate
The sources are credible
The advice is safe
Important limitations are explained
The recommendations can be applied in the real world
Professional input is obtained where necessary
Responsibility is not only an ethical issue.
It also makes good business sense.
If you want to build a profitable, long-term business, you need
people to trust you.
You want customers to feel that your product was worth buying.
You want them to leave positive reviews, recommend you to other
people and buy from you again.
That won’t happen if your products feel like unedited answers
copied directly from ChatGPT.
AI Is a Tool - Not the Publisher
None of this means you shouldn’t use AI.
Quite the opposite.
AI can be an outstanding assistant.
It can help you:
Explore a subject
Generate ideas
Develop an outline
Summarise complex material
Identify gaps in your draft
Improve the structure
Rewrite awkward sentences
Create checklists and examples
Suggest questions your reader may have
But it should assist your judgment, not replace it.
Think of AI as an enthusiastic junior researcher.
It can work remarkably quickly and occasionally produce brilliant
ideas.
But it can also misunderstand the task, overlook important context,
repeat outdated information or confidently invent something that
isn’t true.
You wouldn’t allow a junior researcher to publish a book without
checking their work.
You shouldn’t allow AI to do it either.
Treat Every Claim as Unverified
When reviewing an AI-generated draft, begin with this assumption:
Every factual claim is unverified until I can confirm it.
Pay particular attention to:
Statistics
Named studies
Books and research papers
Medical dosages
Measurements
Historical dates
Legal requirements
Financial rules
Quotes attributed to real people
Claims about named companies or individuals
AI systems can fabricate sources, confuse similar studies or provide
links that don’t support the claim being made.
The information may sound completely convincing.
That is precisely why it needs checking.
How to Check Your AI Research
Start by searching for the original evidence behind any important
claim.
Whenever possible, look for primary sources such as:
Government departments
Universities
Peer-reviewed research
Official statistics
Professional bodies
Regulators
Recognised medical organisations
The original company or organisation involved
Credible secondary sources can also be useful for quotes,
background information and reported events.
These might include Reuters, the BBC, the Financial Times, The
Guardian or a well-established specialist publication.
But do not assume that something is true simply because AI
provides a source.
Open the link.
Read the relevant section.
Check that the source actually supports the claim.
A real link attached to an inaccurate statement does not make the
statement accurate.
A Useful AI Fact-Checking Prompt
You can ask an AI tool to help identify claims that require checking.
Try this:
Review the article below and list every factual claim that needs
verification. For each claim, suggest the most appropriate primary
source and provide a link where possible. Clearly state when you
cannot verify something. Do not invent sources, quotations or links.
This can help you create a fact-checking list.
But remember: AI should help you identify what to check. It should
not be the final judge of whether its own work is accurate.
You could also use:
What are the most common misconceptions about this subject?
Does my draft repeat any myths, oversimplifications or potentially
misleading advice?
And for a more detailed review:
Review the following draft and assess:
Which claims require supporting evidence.
Which statements may be inaccurate, exaggerated or misleading.
Whether the advice is safe and reasonable.
Whether any important risks or limitations have been omitted.
Whether the marketing language could create unrealistic
expectations.
Explain your concerns clearly and recommend what I should verify
using authoritative sources.
Again, treat the result as a second opinion—not proof.
What If Your Sources Disagree?
Conflicting evidence does not necessarily mean you have reached a
dead end.
This is where you start adding genuine value.
Compare the sources and ask:
Which source is closest to the original evidence?
Which organisation has the strongest expertise?
Is the methodology clearly explained?
Is one source more recent?
Are both sources discussing the same group of people or
circumstances?
Is one source presenting an opinion as though it were a fact?
Are the sources genuinely contradicting each other—or referring to
different situations?
If reputable experts disagree, be honest about it.
You could write:
Research in this area is mixed, and experts have reached different
conclusions.
That is far more responsible than selecting whichever claim makes
your product sound most exciting.
It also helps your readers make an informed decision.
Where appropriate, include a list of your main sources at the end of
the article or product.
This demonstrates that you have carried out genuine research and
gives readers the opportunity to investigate further.
Subjects That Require Extra Care
Some subjects carry far greater risks than others.
These include:
Health, medicine and herbal remedies
Mental health, psychology and therapy
Legal advice
Financial advice
Supplements, dosages and drug interactions
Advice concerning children
Advice for elderly or vulnerable adults
Exercise programmes for people with health conditions
Anything involving significant physical or emotional risk
In these areas, unchecked AI-generated information can cause
genuine harm.
If you don’t have the necessary expertise, you may be better
avoiding the subject entirely.
If you do include this type of information, consult authoritative
sources and consider having the content reviewed by a suitably
qualified professional.
You should also use an appropriate disclaimer explaining the limits
of the material.
However, a disclaimer is not a magic shield.
Writing “This is not professional advice” does not excuse inaccurate,
reckless or dangerous content.
The underlying information must still be researched and presented
responsibly.
How to Stop Your Content Sounding Like AI
Accuracy is only half the challenge.
Your product also needs to be worth reading.
Content produced entirely by AI often lacks personality, experience
and genuine insight.
The warning signs can include:
Repetitive sentence structures
Predictable phrases
Generic claims without real examples
Unnecessarily long lists
Advice without practical steps
Excessive summaries and conclusions
A polished but oddly empty tone
Information the reader could find with a quick search
One familiar AI construction is:
“It’s not about X. It’s about Y.”
There is nothing inherently wrong with that sentence.
But when it appears repeatedly, along with dozens of other familiar
AI phrases, the writing starts to feel manufactured.
The most important problem is not that the content “sounds like
AI”.
It is that it offers nothing distinctive.
Add What AI Cannot Provide
Your personal input is what transforms a generic draft into
something useful and original.
Consider adding:
Stories from your own life
Lessons from your experiences
Mistakes you have made
Results you have achieved
Your personal opinion
Examples you discovered yourself
Relevant customer questions
Case studies
Practical observations
Humour
References to books, films or interests you genuinely enjoy
Details relating to your age, work, family or location
Your values and worldview
The advice you believe matters most
For example, thousands of people could ask AI to create an article
about starting an online business.
But only you can explain:
What confused you when you started
Which mistake cost you time or money
What finally made the subject click
Which shortcuts worked
Which shortcuts caused problems
What you would do differently today
That is your advantage.
AI can imitate a human voice.
It cannot live your life for you.
What If You’re Not a Confident Writer?
You don’t need to write beautifully.
Write your story or opinion in straightforward language and then
ask AI to improve the clarity.
For example:
Correct the grammar and improve the flow of the passage below.
Preserve my meaning, personality and informal British tone. Do not
add clichés, exaggerated claims or generic motivational language.
Do not remove personal details or rewrite it in a corporate style.
Always read the revised version carefully.
AI has a habit of polishing away the very quirks that make your
writing sound like you.
If a sentence doesn’t feel like something you would say, change it.
Be Careful With AI-Detection Tools
You may decide to check your finished product using a service such
as Originality.ai.
These tools can help flag passages that appear formulaic or heavily
AI-generated. Some also offer plagiarism and fact-checking
features.
But no AI detector is perfect.
Human writing can be incorrectly identified as AI-generated, while
edited AI content can sometimes pass as human.
So don’t treat an AI-detection score as a certificate of quality.
A low AI score does not prove that your content is accurate,
original, helpful or safe.
The most important tests are still:
Have you verified the facts?
Have you added genuine insight?
Is the advice responsible?
Does the content help the reader?
Are you comfortable putting your name on it?
Your Pre-Publication Checklist
Before publishing any non-fiction content, ask yourself:
✔ Can every important factual claim be traced to a credible source?
✔ Have I opened and checked the sources myself?
✔ Is the advice safe and practical in the real world?
✔ Have I explained important limitations or areas of uncertainty?
✔ Have I included at least one genuine human example, story or
observation?
✔ Does my own voice appear throughout the content?
✔ Have I removed generic, repetitive or padded material?
✔ Would I be comfortable with a knowledgeable professional
checking it?
✔ Have I used an appropriate disclaimer where professional advice
may be required?
✔ Would I feel happy recommending this product to someone I
care about?
That final question may be the most important of all.
The Golden Opportunity
The internet is filling up with cheaply produced AI material.
At first, that may look like bad news for anyone hoping to create
and sell information products.
But I believe it creates an opportunity.
As the amount of generic, unreliable content increases, trustworthy
human-led content becomes more valuable.
People will still pay for information that:
Saves them time
Solves a real problem
Is clearly explained
Has been properly checked
Includes genuine experience
Comes from someone they trust
The winners will not necessarily be the people who use AI to
publish the largest number of products in the shortest time.
They will be the people who use AI intelligently, while adding the
research, judgment, responsibility and personality that the
technology cannot provide by itself.
AI can help you create the product.
But you must remain its editor, publisher and responsible human
being.
The buck still stops with you.
There is much more I could share about creating responsible, high-
quality information products.
So I would be interested to know:
What are you most interested in publishing—ebooks, newsletters,
digital courses or membership content?
And what would you most like help with: research, writing, fact-
checking or the publishing process?
For more free advice go to https://letsimproveyourlife.com/