Screenshot
What would you say if I asked you to buy my box of weight-loss soap?
You would laugh. After all, soap is a topical product. You can’t burn the fat within your body by applying a substance to your skin. Human biology does not work that way.
Any attempts I would make at explaining how the soap penetrates your skin to attack the fat cells beneath would fall on deaf ears. You would rightly dismiss my product as ordinary soap with a flashy package and an eye-watering price tag.
But what if you saw your favourite Ugandan musician on YouTube praising my soap? What if she attested to losing a whopping 40kg using my product? Would that give you pause?
What if a Ugandan top politician appeared on your social media feed singing the same song? Your suspicions would fade. After all, such people would not risk their reputations by selling a fake product to the masses.
Singer/talk show host Kelly Clarkson’s biggest fans probably thought the same thing when the artist began promoting keto gummies for weight loss. Except, it was not her. It was an AI deep fake.
Deep fakes of popular celebrities have become disturbingly common on social media, because AI has become more powerful and much easier to use. Add a few video clips to the right AI tool, and it will create a digital copy of your target, making them do and say whatever you want.
You may argue that AI is not sophisticated enough to create convincing digital humans, and you would spot the fake celebrities instantly. However, the best scammers will show you real footage of Kelly Clarkson, only relying on AI to alter her lips to match the phony audio they have attached to her video.
When was the last time you watched a celebrity’s mouth to determine whether their lips were perfectly synced to their words? Probably never. YouTube is flooded with fake news clips in which ‘news anchors’ highlight financial or medical breakthroughs that ‘experts don’t want you to know about.’
Human beings will believe anything if the right person says it. If you saw BBC News reporters interviewing an ordinary couple who received millions of shillings after sending a message to a said phone number, you may question whether those scammers on WhatsApp who routinely make the same offers to you could be genuine.
You may not realize that someone constructed that entire scene using AI. You can’t rely on platforms like YouTube to help you because Google, which owns YouTube, keeps promoting its own AI tool (Veo3), which allows users to build realistic videos within seconds.
Yes, the platform deletes hundreds of thousands of fake ads annually, but I don’t expect them to make much of an effort to kill this trend when it benefits their AI platform. You must learn to rely on your senses. Pay attention to everything.
Look for obvious signs of fakery. AI gets so many details wrong. Have you seen that video of a dog pulling a stroller with a baby from the edge of a pond? It looks genuine until you take a closer look and realize that the dog is pulling air.
It isn’t even touching the stroller. Listen to the voices. For now, AI cannot perfectly mimic human voices. It gets close. However, the final product is weirdly stiff. You will also notice silly mistakes in the script.
For instance, the fake celebrity will mispronounce obvious words. Additionally, consider the product being advertised. Ask yourself why Oprah Winfrey, of all people, is promoting weight loss soap when she is already so rich. As dangerous as the internet was a month ago, it has become even more dangerous today. Be careful.
katmic200@gmail.com
About the Author
