Technology is rapidly outpacing what the human mind can discern.
Artificial intelligence can fabricate images and videos that look completely real to the average person in the blink of an eye.
However, most people think They’re good at telling the difference between what’s real and what’s not, so put your skills to the test.
Below are six sets of images. In each case, one was real and the other was created by AI. Try it yourself to find out which is which. The answer is at the end of this story.
Here’s what to keep in mind when checking photos, according to the Post’s photo editors: Does the person seem “too sophisticated” for his or her surroundings? Does their face look too symmetrical and perfect? Do their clothing have natural wrinkles, texture, or signs of wear? Are there visible tufts of hair around their head?
John Villasenor, a professor of law and engineering at UCLA, told the Post that he suggests looking for “inconsistencies in lighting and details that don’t really make sense.”
Extensive testing by the Royal Society of Open Science found that a person of normal ability could distinguish between an AI image and a real person only 31 percent of the time.
The study also found that the participants were cheeky and indeed spotted the fakes far more often than they did themselves.
Anatoly Kvitnitsky, CEO and founder of AI or Not, works with companies to find computer-generated images. He also says the perks aren’t necessarily in the face itself.
“The human eye has to look for things in the background. AI is very good at creating a believable main subject, but a person’s face can look blurry in the background. In video, people look stationary.
“If there’s a car in the background, look at the license plate. It might not be perfect. The subheading on the sign might be gibberish. Now AI does a quick job in the background,” he told the Post.
But it may not last long. In the early days of AI, people could easily spot crooked teeth, glasses or accessories that blended into the skin, or ears that weren’t attached properly, but the technology quickly surpassed that. Kvitnitsky says today’s generators create pores and even imperfections.
“There is an arms race between creators and detectors,” Villasenor added. “As creation technology improves, detection technology will try to catch up.”
Kvitnitsky’s company works with clients such as insurance companies to verify the authenticity of images of damaged vehicles, scans of ID cards and checks.
The technology he uses analyzes images at the pixel level to see if they were taken with a real camera.
Images created with publicly available programs such as Google Gemini, Adobe Firefly, and ChatGPT are easiest to find because they have a line embedded in the code that tells you which image generator created them and when.
But if you are not a computer, the odds are against us all the more. A British study published in November 2025 found that even so-called super-recognizers, who are naturally gifted at facial recognition, only have a slight advantage in recognizing human faces 54 percent of the time.
The inundation of computer-generated images in advertising and social media is also subconsciously making people accustomed to seeing AI faces.
If misused, this technology could have serious real-world consequences. In 2024, a Hong Kong financial official was lured into a video conference, apparently attended by his company’s chief financial officer and other colleagues. After being persuaded to transfer $25 million to an external account, the CFO and other employees learned that it was generated by AI. Although the request was fake, the money transferred was legal tender.
Kwitniski believes that this problem will be of great concern to society as a whole in the long run.
“The biggest fear I have with AI is that people will doubt what they are seeing and hearing,” he said.
“Sometimes we see something real and then assume it’s fake. That feeds our prejudices. If you don’t want to believe something, just say: [dismiss] That as AI. ”
Another real-life example emerged last week following the murder of drug lord Nemesio “El Mencho” Oseguera Cervantes by Mexican authorities.
The next day, a photo of a model named Maria Julissa apparently sitting next to him appeared on the Internet, claiming that the two were in a romantic relationship.
Julissa denied knowing or even meeting El Mencho, but it’s easy to see the danger inherent in people linking you to cartel narco-terrorists.
As the lines continue to blur, Kwitniski himself acknowledges that under the right circumstances, it’s possible to be fooled by something AI-generated.
“I have three boys and am the CEO of an AI detection company, but if [something appeared to have happened to them] “They sent me a picture of one of my sons, and my emotions made me forget everything I knew. I just reacted to visual cues,” he admitted.
Answer: 1) B, 2) B, 3) A, 4) B, 5) A, 6) A
