Misinformation spread through news media tripled in three months

Misinformation spread through news media increased in Bangladesh during the July-September quarter of 2026, with the number of cases involving media outlets more than tripling from the previous quarter. 

Fact-checkers identified 76 instances of misinformation spread through news media between July and September, up from 21 in April–June. Though social media still remained the preferred channel for spreading misinformation.

Overall, Bangladesh’s fact-checking organizations documented 1,196 unique misinformation claims between July and September 2026, down 5.5% from 1,266 in the previous quarter. 

Like the previous quarters, politics comprised the largest category of fact-checks in Bangladesh with 647 claims or 54.1% of the total unique claims fact-checked during the quarter.

However, disaster-related fact-checks rose more than sevenfold in the last quarter (July-September), while sports misinformation fact-checks nearly doubled, coinciding with major events such as floods in Bangladesh and Nepal, and the FIFA World Cup.

News media account for growing share of misinformation

The number of fact-checks on misinformation spread through news media increased to 76 in the July-September quarter compared to the 21 in the previous one. Their share of all fact-checked cases also increased from 1.7% to 6.4%.

However, social media still remained the dominant channel, accounting for 90.7% of all fact-checked misinformation during the quarter, although its share declined from 94.9% in the previous quarter. The number of cases spread through social media also fell from 1,201 to 1,085.

Among the three months of the third quarter, July recorded the highest number of misinformation cases spread on social media, while September recorded the highest number spread through news media.

Among the outlets involved, 28 had at least five reports, videos, images or photocards fact-checked. Daily Ittefaq was flagged most often (13 times), followed by Daily Jugantor (12) and Bangla TV (10).

The increase in news media-related misinformation points to recurring problems in reporting and content presentation.

An examination of the cases reveals several recurring patterns in how news media disseminate misinformation, often through the misuse of visual evidence, inaccurate identification, misleading contextualization, and unverified reporting. 

One major pattern is the misuse of photographs and videos ( 1, 2, 3, 4, 5, 6, 7, 8), where old, unrelated, or AI-generated ( 1, 2, 3, 4) visuals are presented as evidence of recent events, including natural disasters, military attacks, accidents, and alleged rescue operations. In August, Dismislab fact-checked a Daily Inqilab report about a boat accident in Nigeria that used an image from an unrelated incident in Zimbabwe.

Another pattern of misinformation is misidentification and false attribution, in which media outlets use the wrong person’s photograph, confuse individuals with similar names, or incorrectly identify someone (1, 2, 3, 4). A third pattern involves communal framing, often by Indian news outlets, which portray incidents in Bangladesh as religious persecution without enough evidence ( 1, 2, 3, 4). 

A closer look at the 76 cases shows where the misinformation came from. International events accounted for 32 cases, domestic matters for 29, sports for 11, and four were random clips. Disasters made up most of the international cases (22 of 32), and video was the dominant format, used in 18 of the 32.

Domestic cases were more varied. Graphic cards were the most common format (10 of 29), followed by text (7) and video (6). Indian outlets were behind seven of these cases, four concerning geopolitics and three on religion. Bangladeshi outlets accounted for the other 22, led by general misinformation (12) and politics (six). The four random clips, which sit outside the 29 domestic cases, were all AI-generated videos.

In case of sports, Bangladeshi outlets spread 10 of the 11 misinformation cases: eight related to the FIFA World Cup and two to cricket. The other case was spread by an Indian outlet.

Disaster related misinformation rose sharply amid major floods

Disaster-related fact-checks increased from eight unique claims in April–June to 60 in the July–September quarter. Its share of all unique claims rose from 0.6% to 5.0%, making it the quarter’s fastest-growing category, with July recording the highest number of disaster-related misinformation. 

Heavy rainfall began in Chattogram on 5 July 2026, triggering flooding across the region. According to a 12 July report by the Ministry of Disaster Management and Relief, floodwaters had inundated 59 upazilas across seven districts. Starting from July 10, the misinformation content appeared primarily as images and videos and included AI-generated images and videos, as well as old footage from Pakistan and Turkey, presenting it in the context of the ongoing floods. Much of the content depicted women and children in distress.

Another major event contributing to disaster-related misinformation was the flash flooding and landslides in Nepal on 26 August. Misinformation related to the incident began circulating in Bangladesh on 27 August. A similar pattern emerged, with AI-generated images and videos, alongside old footage from unrelated events in India, Pakistan, the Philippines, Vietnam, China, the United States, Japan, and Turkey, being used to misrepresent the disaster in Nepal.

Sports fact-checks nearly doubled during World Cup

Sports-related fact-checks increased from 38 unique claims in April–June to 71 in July–September, rising from 3% to 6% of total debunked claims compared to the previous quarter.

The increase was concentrated heavily during the FIFA World Cup in July, when fact-checkers recorded 62 sports-related claims. The number fell to eight in August and one in September.

Sports-related misinformation circulated primarily through images, videos, and photocards on social media and in news media. Prominent footballers featured in the misleading content included Neymar, Erling Haaland, Cristiano Ronaldo, Lionel Messi, and Lamine Yamal. Examples include AI-generated images and videos depicting Lamine Yamal holding a Palestinian flag after the final, as well as AI-generated videos featuring Erling Haaland. The misinformation also included misleading content showing the Argentine national team holding the Bangladeshi flag and fake quotes attributed to Neymar and Lionel Messi, and claims that Yamal had refused to shake hands with Donald Trump during the final. 

Politics remained dominant theme, but its share dropped

Political misinformation accounted for 647 unique claims in July–September, down from 718 in April–June. Its share of all claims also fell, from 56.7% to 54.1%. Despite the decline, politics remained by far the largest category, accounting for more than half of the misinformation tracked during the quarter.

August recorded the highest number of political misinformation cases during the quarter, with fact-checkers documenting 212 in July, 258 in August and 177 in September. August’s increase coincided with the second anniversary of the July mass uprising, observed nationwide on 5 August.

Following the initial spread of misinformation on June 26 claiming that Chhatra League activists’ bodies had been recovered from the Turag River, a similar claim resurfaced in July. Unrelated body-recovery footage was falsely presented as showing the recovery of another Awami League activist’s body in connection with the so-called “Turag Tragedy” (1, 2, 3, 4).

Also in July, several pieces of misinformation circulated after a private video of Satkhira-4 MP Gazi Nazrul Islam went viral, including false claims that the footage was AI-generated. False reports also circulated on social media and in online news outlets claiming that Gazi Nazrul Islam had been arrested. These reports confused him with Mohammad Nazrul Islam, a former Satkhira District Council chairman and Awami League leader who had actually been arrested.

As in the previous quarter, misinformation during July–September largely centred on Sheikh Hasina’s alleged public appearances and possible return to Bangladesh. Old footage of her meetings and events with various leaders and presidents was recirculated as recent. Meanwhile, videos and images of Awami League processions in different districts were circulated, with AI used to exaggerate crowd sizes (1, 2).

Another major misinformation narrative emerged around the National Citizen Party’s (NCP) “July March”. On 28 July 2026, an NCP convoy was attacked and vandalised during the party’s march in Habiganj. The NCP blamed local activists of the Bangladesh Nationalist Party (BNP) and Jatiyatabadi Chhatra Dal (JCD), while both groups denied involvement and attributed the incident to angry local residents. 

NCP activist Alif Chowdhury reportedly lost vision in one eye in the incident. At least 10 pieces of misinformation of various types circulated in connection with the attack. Some directly targeted Alif. These included a false claim that the Jamaat-e-Islami amir had visited him in hospital, footage that misidentified another person as Alif being attacked by BNP activists, and claims that Alif was faking his vision loss.

Religion and international affairs-related misinformation decline

Misinformation verification related to religion and international affairs followed a similar decline between the second and third quarters. International affairs-related claims fell from 10% to 9% of all checked misinformation, while religion-related claims decreased from 8% to 7%. 

July recorded the highest number of international affairs-related misinformation cases in the third quarter, with the Indian Cockroach Party protest emerging as a notable topic. In contrast, religion-related misinformation peaked in September.

Religion-related fact-checks showed that misinformation during the quarter followed different narratives in Bangladesh and India. In Bangladesh, much of the misinformation framed Hindu communities as victims of violence, persecution, or mistreatment. Examples included the recirculation of videos of vandalism with a misleading religious-persecution narrative (1, 2, 3, 4, 5), a false claim that three Hindus had been killed in a bomb blast in Sitakunda, and Indian media reports that misrepresented videos as showing a Hindu student being targeted at Jahangirnagar University or Hindu homes being set on fire (1, 2, 3). 

In contrast, religion-related misinformation about India often portrayed Muslims, particularly women, as targets of harassment, abduction, or violence by Hindus. For example, a video claimed to show Hindu men abducting a Muslim girl ( 1, 2, 3), Muslims as victims of violence by Hindus (1), and illegal Bangladeshi Muslims have raped Hindu women (1). 

Of the international affairs-related claims debunked during the quarter, 31 centered around protests in India associated with the Cockroach Janta Party (CJP). Examples include fabricated videos of protesters being arrested ( 1, 2, 3, 4), AI-generated footage falsely showing Narendra Modi fleeing demonstrators (1, 2), and misleading claims involving celebrity support for protest movements ( 1, 2, 3).

Law-and-order misinformation rises

Law-and-order misinformation rose from 104 unique claims in April–June to 129 in July–September, an increase of 24%. The category became the second-largest after politics, accounting for 10.8% of all unique claims. July recorded the highest number (51) of law and order-related fact-checks during the quarter. 

These types of misinformation frequently involved allegations of murder, rape, robbery, mugging, and violent physical assaults. Several cases involved incidents from other countries, particularly India, falsely presented as having occurred in Bangladesh. For example, videos showing a woman being attacked with a sharp weapon in Maharashtra and a man being murdered in public in India were circulated as incidents in Bangladesh. Similarly, footage showing the recovery of a teenage girl’s body in West Bengal was misrepresented as evidence of rape and murder in Bangladesh. 

Misinformation also distorted details of actual incidents, including the discovery of the bodies of former teacher Ganapati Chakraborty and his family in Rangpur.

Older incidents, AI-generated content, and footage of unrelated events were also misrepresented as recent developments in Bangladesh. For example, a video recorded during the July 2024 mass uprising in Mawna, Gazipur, was recirculated as footage of recent clashes between garment workers and police in Sreepur. Similarly, an AI-generated video of police violence was shared to support claims of police brutality before 5 August. Among other cases, a video of a fire in Myanmar was falsely presented as footage of airstrikes near Bangladesh’s Teknaf border. Furthermore, claims about bomb blasts, terrorist attacks, and militant activities ( 1, 2, 3)  in Dhaka were found to be misleading, with images and objects misrepresented and unrelated to the alleged incidents.

A notable surge in fact-checks also centered around protests by Higher Secondary Certificate (HSC) examinees. Examples included an old photograph from a 2025 workers’ protest in Nilphamari, falsely presented as evidence that police had shot and killed two students during an HSC protest. Videos from the July 2024 mass uprising and an April procession by Bangladesh Islami Chhatra Shibir were also recirculated as footage of HSC examinees protesting. 

Several staged videos depicting shootings, armed threats and extortion were falsely circulated as real incidents amid concerns about Bangladesh’s law-and-order situation. These included videos of armed men in a market, a truck driver’s staged abduction, and an alleged extortion incident at Jamtail railway station in Sirajganj. Footage from movie shoots in Assam and Patna, India, was also misrepresented as showing the Arakan Army attacking Indigenous people in Rangamati and armed individuals shooting civilians during Bangladesh’s 5 August 2024 uprising.

Video remains the format of choice

Videos were the most common format among the claims fact-checked in July–September. Fact-checkers recorded 617 unique video-based claims, accounting for 51.6% of the total. This was slightly higher than the 613 claims, or 48.4%, recorded in April–June.

Graphic cards remained the second-most common format, although their number fell from 387 to 347. Image-based claims declined from 188 to 164. In contrast, text-based claims increased from 22 to 42.

The continued prominence of video reflects the recurring use of old footage, clips from unrelated countries and events, and AI-generated videos to support misleading narratives. Graphic cards also remained a prominent tool, particularly in political misinformation, where fabricated quotations and false reports were attributed to established media outlets.

AI-related misinformation declined in absolute numbers. The collected data flagged 163 unique claims as involving AI in July–September, down from 191 in April–June.

However, AI-generated or altered content remained visible across several categories, including politics, sports and disasters. The examples from the quarter show that the use of AI did not depend on a single topic: it appeared in fabricated political imagery, false depictions of sporting events and invented scenes from natural disasters.