There was a time when it was easier to believe things because there was no use of technology like artificial intelligence. But in this era of the 21st century, things have changed. Any event or information can not be believed very easily, and even if it is believed, it turns out later that it is a false report or news created using artificial intelligence. The tools generating fake images and videos have gotten good enough that looking closely barely tells you anything most often. Fake videos, images, and news created by AI are problems all across the world.
The European Parliamentary Research Service has estimated that the volume of deepfake video circulating online could rise from roughly 500,000 clips in 2023 to around 8 million by 2025, a sixteen-fold jump in just two years. This problem predates AI, and AI fakes spread so fast. MIT researchers tracked about 126,000 verified true and false stories across 4.5 million tweets in the largest study of its kind, and found false stories got reshared 70 percent more often and hit their first 1,500 people six times faster than true ones. The researchers found this was not driven by bots. It was ordinary people sharing false stories faster than true ones, mainly because false stories tend to be more novel and more emotionally charged.
Some AI fakes worth mentioning. After Hurricane Helene hit the US in 2024, AI-generated images flooded social media, including one widely shared photo of a crying child clutching a puppy in a rescue boat, floating through the wreckage. The image was entirely synthetic, yet it spread faster than official emergency updates. During Hurricane Milton the same year, numerous misleading videos went viral, including the one that was actually old footage from 2018 with a tornado artificially added to it. In September 2025, an AI-generated video showing the CN Tower in Toronto engulfed in flames was viewed more than twenty million times on Facebook before the tower’s own management had to publicly confirm ”no fire”.
Since early July, the catastrophic flooding across upazilas like Satkania, Banshkhali, Chandanaish, and Lohagara in Chattogram, Bangladesh, has left more than seven hundred thousand people marooned, with many lives lost. That is a real brutal, ongoing humanitarian crisis. But alongside this genuine disaster, a parallel “information disaster” was unfolding on social media. One video showed a CNG auto-rickshaw being carried on bamboo planks across a flooded canal, with a caption claiming rescuers were saving a pregnant woman trapped inside. The actual footage was from Bandarban’s Naikhongchhari, and there was no pregnant woman inside. It was just an emotionally charged caption stitched onto real footage of a real rescue.
A few days ago on Facebook, there was a striking video — a pregnant woman clinging to a tree branch in floodwater, fighting to survive. The caption said a helpless mother, pregnant, was trying to save herself and her unborn child. It was emotional. But the whole thing was fake. The original footage was from 2024 — a woman posing on a tree branch during a leisure trip to Tanguar Haor in Sunamganj. Someone had used AI to dress it up as a scene from the Chattogram floods. But the fabrication was caught by the unnatural shape of the woman’s fingers gripping the branch, the exact spot where AI image generators still routinely stumble.
Fabricated images lead to not only disinformation and misinformation but also disbelief in real content. After Cyclone Ditwah hit Sri Lanka, a real photo of bodies piled along a road in Gampola went viral, and people assumed it had to be AI-generated. Multiple forensic checks confirmed the photo was genuine, and police officially verified the victims were local residents. Legal scholars call this the “liar’s dividend”. Alternatively saying, once people know fakes exist, they become more willing to dismiss genuine evidence as fake, or hesitate when a real warning is issued. In many situations, including a disaster, that hesitation can cost lives.
But it is often difficult to differentiate real content. A Pew Research Center survey from October 2025 found that 52 percent of Americans find it difficult to tell what is true and what is not when they get news about elections. So, how can real content be separated from fake panic? There are several ways. It is vital to check whether any claim has been verified by a credible national or international media outlet, district administration, or disaster management ministry. Moreover, if an image or video evokes a strong emotional response like fear, tears, or sudden anger, it is helpful to understand the context before sharing it. What is the scope of the incident? Content creators plan to create this intensity.
There are also several other ways. The use of reverse image search tools like Google Lens or TinyEye can help. These can quickly reveal whether the image has previously appeared online in a different context. Moreover, AI-generated images still have trouble capturing small details such as hands, fingers, teeth, and background objects, exactly the kind of error that gave away the viral video image from Chittagong, where one of the woman’s hands was completely missing.
It is not that actions are not taken at all. Legal actions are taken in many countries. In Zhejiang province, China, for instance, two people were given punishment with over a year in prison after their media company used AI to fabricate a video of a massive fire engulfing an industrial building, purely to generate traffic and ad revenue. But legal actions are meagre, even if they are needed many times. Moreover, it is often difficult to find those who are involved in creating such videos or images and take action. But these need to be made stronger.
Independent fact-checking plays a role. During the LA wildfires in January 2025, AI-generated clips showing the Palisades Fire were circulated with captions asserting a specific and dramatic death toll, but independent detection analysts confirmed the footage itself was synthetic and generated to accompany fabricated casualty figures. Television fact-check programs can also help fight fake video. But emphasis needs to be given to creating awareness and promoting verification of the veracity of any social media video, image, or news after seeing it.
DISCLAIMER: The views expressed are those of the author(s) and do not necessarily reflect the official policy or position of the Magazine and its editorial team. The views published are the sole responsibility of the author(s).
