How to Spot AI Content: Simple Ways to Detect Fake Text and Images
By Rob Budd | 20+ years driving growth for top UK brands
Ever read something on LinkedIn and thought, “I know what the next paragraph is going to say”?
Not because you've seen that exact post before. More because you've seen the same type of post hundreds of times.
It starts with a slightly dramatic observation. Then comes a short sentence for emphasis. A few perfectly balanced paragraphs follow, usually ending with three bullet points and a neat little lesson about the future.
You can't always prove it was written by AI. But increasingly, you can feel it.
The same is happening with images. Scroll through social media and you'll see them everywhere: impossibly perfect people, strange lighting, backgrounds that don't quite make sense and images that look convincing for about two seconds, until you actually look at them.
We've already written about why modern advertising has gone completely beige, and AI is definitely part of the problem. Not because AI itself is inherently bad, but because everyone is using the same tools in roughly the same way.
The result is a lot of content that looks polished, but says very little.
So, what should you look for?
What actually counts as AI content?
Put simply, it's content generated wholly or partly using AI tools.
That could be a blog post, a LinkedIn update, an advert, an image or a video. Sometimes the AI has done everything. Other times, a human has used it as a starting point and then properly edited the result.
That distinction matters.
Using AI to help research a subject, organise your thoughts or get past a blank page is one thing. Pressing a button, copying the output and publishing it without adding anything of your own is something else entirely.
And, more often than not, you can tell.
Three signs that text might have been written by AI
1. It all sounds a bit too perfect
This is probably the biggest giveaway.
AI-generated content often has a very even rhythm. Every paragraph is neatly structured. Every sentence is complete. Everything flows perfectly from one point to the next.
Real people don't always write like that.
Sometimes we go off on a tangent. Sometimes we make a point and move on without explaining it to death. We use short sentences when we're annoyed. Longer ones when we're trying to explain something properly.
Human writing has a bit more variation to it.
AI writing can sometimes feel like listening to someone who has rehearsed every sentence before opening their mouth.
2. It uses a lot of words without saying much
You've probably seen the usual suspects: delve, landscape, game-changer, testament, tapestry, furthermore and, inevitably, in conclusion.
The problem isn't the individual words. People use them too.
It's when the writing starts sounding like it has been assembled from a collection of phrases designed to make it sound intelligent.
You get paragraphs that appear to be making an important point, but by the end of them you're not entirely sure what that point was.
That's often a sign that nobody with a real opinion has gone back and asked: "Would I actually say this?"
3. There's no experience behind it
This is the one I notice most.
Generic AI content can explain almost anything. What it can't do, at least not truthfully, is tell you what happened when something went wrong for a client, what somebody learned from a failed campaign or why an experienced marketer disagrees with the conventional advice.
That's where the interesting stuff usually is.
If an article is entirely made up of broad statements, generic advice and things everyone already agrees with, there's a good chance it hasn't had much human input.
A useful test is this: could anyone have written it?
If the answer is yes, it probably needs more of the person whose name is attached to it.
How to spot AI-generated images
Images are getting harder to identify.
The obvious examples of six fingers and unreadable signs aren't as reliable as they once were. AI image tools have improved enormously, and some generated images are now difficult to distinguish from real photography at first glance.
But there are still things worth looking for.
1. Look beyond the main subject
The person or product in the middle of the image might look completely convincing.
The background is often where things start to get strange.
Look at objects that aren't the focus of the image. Signs, furniture, hands in the background, reflections, jewellery, glasses and anything with fine detail can still reveal inconsistencies.
Sometimes nothing is obviously wrong. It just doesn't quite make sense.
2. Ask whether the lighting makes sense
AI images can look incredibly polished, but polish isn't the same as realism.
Shadows might fall in slightly different directions. Reflections might not match the scene. A face can be perfectly lit while the rest of the image appears to be illuminated from somewhere else.
Again, you don't always spot it immediately. Often it's just a feeling that something is off.
And that's usually enough to make you take a second look.
3. Watch for details that change when you inspect them
Patterns on clothing, text, jewellery, glasses and objects people are holding can all behave strangely in generated images.
You might see something that looks perfectly normal until you zoom in. Then a button disappears, a pattern changes halfway across a shirt or a pair of glasses suddenly stops looking like an actual pair of glasses.
The technology is improving quickly, though, so I wouldn't rely on any single "AI giveaway". The best approach is to look at the image as a whole and ask whether the details genuinely hold together.
Has AI spam disappeared, or has it just got better?
That's the more interesting question.
We've previously asked whether AI ad slop has vanished from social feeds or has it just got smarter.
I don't think the problem has gone away.
I think people have simply become better at hiding it.
The better examples of AI-assisted work don't look like obvious AI content because someone has actually put some effort into them. They've used the technology to speed up research or develop an idea, then rewritten it, checked it and added something of their own.
The worst examples are still easy to spot. They're the endless stream of generic articles, LinkedIn posts and adverts that all seem to have been produced from the same prompt.
Different companies. Same voice.
Does it actually matter?
I think it does.
Not because every piece of content has to be handcrafted from scratch, and not because using AI is somehow cheating.
It matters because your marketing is supposed to give people a reason to choose you.
If your website sounds exactly like every other website in your industry, you're making that job harder.
People notice when you've put no effort in
Your potential customers might not immediately say, "This was written by AI."
But they can still sense when something feels generic.
If your marketing is full of vague claims, recycled advice and personality-free content, it doesn't give people much reason to believe you're any different from the next company.
We've already looked at the backlash against corporate AI spam and why authentic content wins.
The issue isn't really whether a machine was involved in producing the content.
The issue is whether there is anything worth saying once it's finished.
Search engines care about useful content, not just more content
Publishing 200 articles that all say roughly the same thing isn't a content strategy.
Neither is asking AI to generate endless pages targeting slightly different keyword variations and hoping Google rewards you for volume.
The web already has enough content.
What people are looking for is something useful, accurate and relevant. Something written by somebody who actually knows what they're talking about, ideally with experience to back it up.
You can't stand out by copying the same machine as everyone else
This, for me, is the biggest issue.
If ten companies all ask the same AI tool to write about the same subject, they're probably going to get ten versions of the same answer.
The wording might be different, but the ideas will often be very similar.
That's not how you build a distinctive brand.
A strong brand has opinions. It has a point of view. Sometimes it says things that competitors wouldn't say. Sometimes people disagree with it.
That's fine.
In fact, it's usually more interesting than another perfectly optimised article telling everyone to "embrace innovation in today's rapidly evolving landscape".
The sensible way to use AI
I'm not anti-AI.
We use AI tools. Most good marketers probably do.
They're useful for research, brainstorming, analysing information and getting past the blank-page problem. Used properly, they can save a huge amount of time.
But they should be a starting point, not a substitute for thinking.
The final version still needs a person to decide what's worth saying, what should be cut, where the opinion is and whether the content actually sounds like the brand behind it.
Because the more AI-generated content floods the internet, the more valuable genuine experience and an actual point of view become.
At Axcellerate, we use technology where it makes sense. But we don't think your brand should sound like it was assembled by an autocomplete function.
If your marketing has started to sound like everyone else's, maybe it's time to put a bit more of you back into it.