It makes life simpler, but it’s also being used to scam others under the guise of ‘doing good’. We’ve shared a few helpful insights about how to spot AI. Not to break your trust, but to protect it.
South Africa (03 November 2025) – Did you know that ChatGPT has around 700 million users every week? Sam Altman, the founder of the company behind the generative AI chatbot, has estimated that 10% of the overall population on the planet are ChatGPT users.
Let that sink in.
It’s used by everyday people for everyday things. A recent study by OpenAI and Harvard economist David Deming, the largest ever on ChatGPT use, found that three-quarters of all conversations on ChatGPT are about practical guidance like asking for advice, writing something, planning, or getting quick information.
How we use AI can also be thought of in terms of asking, doing, and expressing.
“About half of messages (49%) are ‘Asking,’ a growing and highly rated category that shows people value ChatGPT most as an advisor rather than only for task completion. Doing (40% of usage, including about one third of use for work) encompasses task-oriented interactions such as drafting text, planning, or programming, where the model is enlisted to generate outputs or complete practical work. Expressing (11% of usage) captures uses that are neither asking nor doing, usually involving personal reflection, exploration, and play,” shares OpenAI.
It’s become a really helpful tool that makes life easier. But there’s always a downside. Or, downsides.
There are ongoing arguments about the risk that AI may pose. Will it take over jobs, will the reliance of it eventually dim our ability to think for ourselves, and more existentially, could AI eventually become a form of super-intelligence that acts against humanity’s best interest?
It reaches deep. One of the most immediate concerns to us, though, is people using good tech for bad things.
Like we saw last week, an AI-generated crowdfunding campaign could have easily landed unsuspecting victims in its web of lies. It faked a story designed to play on people’s feelings and urged them to donate towards ‘the cause’. A scam set up especially to target compassionate people who want to help others.
Related Stories
St John's Choir 'Heals the World' with Michael Jackson Performance
Dear Tired Humans, You’re Actually Doing Okay!That is downright wrong.
That’s where discernment comes in. It’s not about being cynical but rather about taking a minute to check that it all adds up. That line of questioning might be enough to protect your kindness and your trust.
In the spirit of learning how to ‘separate the good from the Grok’, as our Editor says, here are a few tips to help you spot whether something is AI or not.
We’ve separated according to three different kinds of content – video, image and text.
Video That Seems Uncanny
AI-generated videos, or deepfakes, are getting scarily realistic. But there’s often something that feels slightly off or uncanny. It’s easy to spot with close attention. It could be in the way someone blinks, or how their smile doesn’t quite reach their eyes. Pay close attention to movement. Real people move imperfectly, and AI hasn’t fully mastered that.
Voices can be another giveaway. AI voices have a smoothness to them. It’s a bit like listening to a voice note that’s been run through a robot filter – you can’t always explain why it sounds funny, but it does.
Lighting and background details also tend to escape AI. Shadows might fall a bit strangely, or light moves in a way that doesn’t add up. If a clip feels unnaturally flawless, that’s your cue to take a second look.
What to look for:
- Eyes that blink oddly or not at all
- Lips that don’t match the words
- Hair or jewellery that glitches or flickers
- Emotion that doesn’t fit the voice
Images That Just Don’t Sit Right
AI images can look hyperrealistic, polished and unusually smooth at first glance. Zooming in is where things get weird fast. Fingers blur together or multiply. Text on signs turns into gibberish. Reflections don’t match, and backgrounds melt or repeat the same pattern.
That gut feeling you get is worth trusting.
What to look for:
- Hands or fingers that look off
- Text that’s unreadable or distorted
- Repeated or ‘melted’ background details
- Shadows or reflections that don’t line up
- Faces that look too flawless or smooth
Word Soup That Reads Too Well
AI can write well, sometimes a little too well. An article by Medium joked that one word, ‘delve’, might be the biggest clue that something was written by ChatGPT.
don’t mind me just doing some delving today
— ChatGPT (@ChatGPTapp) April 17, 2024
People have already compiled lists of AI buzzwords that make it easy to spot.
Different models may tend to overuse certain words. “Grit”, “testament”, “tapestry” and “underscore” immediately raise the red flag. Then there are phrases like “delve into”, “at its core…”, “that being said”, and “to put it simply” (or some variations thereof) that pop up everywhere.
When you stack too many of them together, it starts to feel really uncomfortable. Human writing isn’t perfect, and that’s what makes it good and gives it voice. We use contractions, emotion, and rhythm. We say too much, then delete half of it. We make typos!
So when a piece of writing feels a little too polished, that’s when you cast a closer eye.
What to look for:
- Polished grammar, but vague meaning
- Buzzwords
- Overuse of adverbs
- Phrases that sound like filler
- Tone that jumps too much
- No clear opinion or emotion
The quick gut-check
When something feels off, slow down. Do a quick reverse image search, listen again with your eyes closed, ask a friend or family member to have a look, or confront the source directly. And if you’re still unsure, run it through AI. Ironically, it’s really good at spotting its own work.
We don’t want to villainise AI. It’s a mirror of the people using it. It can make life simpler, or it can twist the truth. It’s up to us to use it responsibly and hone in on being more careful about the things we see and read.
by 
