Nepali Society Trapped By Algorithms, AI Further Spreading Misleading Information
पुस १, २०८२ १५:३२
Kathmandu: The misleading post or video linking the Gen-Z rebellion to the alleged discovery of 32 skeletons inside the Parliament building may have angered you. When you saw photos showing then Prime Minister KP Sharma Oli and Home Minister Ramesh Lekhak in handcuffs, did you wonder, “What will happen now?” or think, “This is wrong”? Whether it was the heartbreaking image of Jasmine, who was claimed to be suffering from bone marrow cancer, or the rumor that Kathmandu Metropolitan City Mayor Balendra Shah (Balen) was wearing black clothes and had not shaved his hair and beard during a surgery, many people have been emotionally affected by such content.
If so, you may have fallen, knowingly or unknowingly, into the trap of a serious digital phenomenon. With the rapid spread of information technology, the internet, and smartphones, distinguishing truth from illusion has become increasingly difficult. By misusing AI tools, some Nepali Facebook pages are turning human emotions into content and exploiting them for reach.
A clear example of this emerged during the Gen-Z protest on 9 September. On that day, amid government repression and rising public anger, YouTuber Tank Dahal posted a sensational claim on social media. He said police had detained 32 protesters inside the Parliament building and expressed fear that the youths might disappear. On the first day of the protest, 19 people had already been killed during clashes. In this context, Dahal cited eyewitness accounts and made a highly provocative claim.
The claim spread rapidly. By the next day, the unverified information that “32 people were detained inside the Parliament building” had morphed into a false claim that “32 skeletons were found inside the Parliament building.”
YouTubers including Bhagya Neupane and Sujan Dhakal presented the claim in an even more emotional and sensational manner. While protesters had damaged parts of the Parliament building and set some structures on fire, the claim that 32 people were burned alive or that skeletons were found was entirely baseless.
After fear and anger had already spread, Nepal Police issued a clarification stating the information was false. A fact-check by TechPana also confirmed that no bodies were found inside the Parliament building. However, by the time the truth emerged, many people had already accepted the false narrative as fact. This incident shows how easily a single piece of information can be distorted and weaponized.
…
The political sphere also became a victim of this trend. An ‘Inquiry Commission on the Incidents of September 8 and 9, 2025 was formed to investigate state repression during the Gen-Z protests and the subsequent vandalism, arson, and looting. Since the commission’s formation, mainstream media reported that it was considering calling then Prime Minister KP Sharma Oli and Home Minister Ramesh Lekhak for statements.
As public anger grew, various groups demanded the arrest of Oli and Lekhak, holding them responsible for the repression. To capitalize on this sentiment, pages such as Laligurans Media, Sudan Gurung, and Jwala Sangraula published sensational posts claiming the two leaders were about to be arrested. Soon, the rumor escalated into claims that they had already been arrested.
Facebook pages with pseudonyms such as ‘Miss Pub 41’, ‘KTM Post’, and ‘Comrade Janu’ used AI to generate fake images showing Oli and Lekhak in handcuffs and police custody. Nepali text in these AI-generated images was often distorted. To make them appear authentic, page operators edited the text manually to mask the flaws.
A political discussion was thus transformed into a “confirmed event” by social media. Some pages even solicited reactions from users, asking what punishment should be given to the “arrested” leaders.
Media researcher and fact-checker Deepak Adhikari says this is a textbook case of misinformation turning into disinformation. He explains that confirmation bias plays a major role. People tend to believe and share content that aligns with their existing beliefs, without verifying facts.
According to Adhikari, discussion itself can be news. For example, discussing whether leaders might be summoned for questioning is news. But discussion does not mean arrest. Pages chasing reach and engagement present discussion as a concluded action.
“People who already want KP Oli and Ramesh Lekhak to be arrested accept fake arrest news as truth and share it,” Adhikari said. “This is similar to how AI-generated images of Donald Trump being arrested went viral globally. A small seed of information is exaggerated and pushed from misinformation to deliberate disinformation.”
Adhikari warns that such AI-generated content not only deceives individuals but also risks destabilizing society. The race for reach is not limited to politics. It also exploits moments of human tragedy.
The case of four-year-old Jasmine Basnet, who was undergoing treatment for bone marrow cancer, is a stark example. YouTuber Bhojraj Thapa had been sharing verified updates while raising funds for her treatment. However, other pages used AI to create fake images and videos showing Thapa and Jasmine’s mother crying and portraying the child as being in critical condition. These visuals went viral and emotionally manipulated thousands, even as the real treatment process was ongoing.
Similar insensitivity was seen in the case of comedian Sitaram Kattel (Dhurmus). While facing difficulties related to a stadium construction campaign in Chitwan, rumors spread after he traveled to the US with his family. Kattel later clarified that his daughter required medical treatment unavailable in Nepal and India. Soon after, AI-generated images showing his daughter undergoing treatment, with Dhurmus and his wife Kunjana Ghimire crying, were circulated widely.
This trend also surfaced during the mourning period of Kathmandu Mayor Balendra Shah. After the death of his father, Dr. Ram Narayan Shah, social media fixated on Balen’s attire, hair, and beard. Some pages circulated AI-generated images of Balen wearing black clothes with long hair and beard during mourning rituals. The confusion was cleared only after his wife, Sabina Kafle, released an authentic photo on the 12th day. By then, AI-generated misinformation had already spread widely.
…
Pratyush Ranjan, Fact Check Editor at Press Trust of India (PTI), also attributes this to confirmation bias. He says people believe what they want to be true. “Old videos from Nepal’s Gen-Z protests are repackaged and shared as incidents from Bihar or other places. Clips taken out of context only increase anger and confusion,” he said, adding that timely dissemination of accurate information and stronger digital literacy are essential.
Kathmandu University associate professor and media scholar Sudhanshu Dahal says the situation is becoming increasingly complex. AI tools are now in the hands of people with superficial understanding but strong intent to manipulate. Traditional media norms are being replaced by social media pages, where anyone can generate realistic-looking content.
“People create such content casually, without considering its consequences. Ordinary users accept it as absolute truth,” Dahal said. “Social media algorithms add fuel to the fire.”
According to him, algorithms trap users in filter bubbles. “If you consume one type of content, the algorithm keeps feeding you the same. You begin to believe that what you see is the entire reality.”
Dahal notes that the problem is not limited to uneducated users. Even educated and influential individuals fall prey to misinformation. He cited an example where a misleading post claimed that a ‘Shivaji figure was seen in the Himalayas’. Despite being entirely unscientific, even a senior university professor shared it. “When critical thinking is lacking even among intellectuals, the situation of ordinary citizens is easy to imagine,” he said.
Such misinformation affects mental health, social harmony, and public security. More dangerously, it fuels mob mentality. Dahal cited a school in Chitwan where students reportedly threatened violence after being influenced by social media narratives. “Creating the belief that mob pressure guarantees outcomes is a serious threat to democracy and the rule of law,” he warned.
Which pages are spreading such content, and why is it spreading?
Analysis shows that many of these posts originate from similar types of pages, including ‘Miss Pabi 41’, ‘Comrade Jamu’, ‘Nabaraj BK’, ‘Givina Magar’, ‘Yubin Magar’, ‘Miraj Dhungana Naamak Group’, ‘KTM Post’, ‘Divya Malla Bam’, ‘Bojraj Thapa’ and others. These pages frequently share AI-generated visuals with similar patterns and narratives.
It appears that unverified information circulating on social media is being deliberately turned into misinformation. These pages and groups are seen spreading confusion by linking human emotions and political issues. Facebook pages and groups named Miss Pabi 41, Givina Magar, Miraj Dhungana, and Bhojraj Thapa have been found misusing the names and photos of individuals who are currently in the public spotlight. Under the guise of fan clubs, such pages are spreading misleading and false information. An analysis of their locations shows that most of these pages are being operated from outside Nepal.
To increase engagement, emotionally charged text is commonly used in these posts. For example, in reference to Balen, captions such as “Even after 12 days of his father’s death, Balen’s tears have not stopped, give one like to show support” are used. Other examples include, “Our daughter is undergoing treatment, everyone please write ‘Get well soon’ for her, those with a heart will like and comment,” “Dear daughter Jasmine is undergoing her final check-up, you will recover soon, do not worry,” “Treatment of Dhurmus Suntali’s daughter has begun, everyone please write ‘Get well soon’,” “Daughter Jasmine’s condition has become more critical, those with humanity please wish her a speedy recovery,” and “Finally, Oli and Lekhak caught in a legal trap, comment whether arresting big fish is right or wrong.”
According to AI expert and Madan Bhandari University of Science and Technology professor Suresh Manandhar, such wording is used as a strategy to play with users’ emotions, increase engagement, and mislead social media algorithms into boosting the post.
“The main thing social media algorithms look at is engagement,” Professor Manandhar said. “They work on a recommendation system. If you and I, or multiple connected users, consume similar types of posts, the algorithm will show you posts I have seen but you have not. When taken at scale, if a post goes viral, it is shown to many users, including those who have not seen it yet.” He explained that because social media algorithms prioritize engagement over whether content is misleading or false, posts published by such pages are more likely to go viral.
पछिल्लो अध्यावधिक: पुस १, २०८२ १५:३२
