Nepal Urges Meta, TikTok to Crack Down on Flood Misinformation and AI-Generated Content

KATHMANDU – The Nepali government has demanded urgent action from social media giants Meta and TikTok to remove graphic imagery, artificial intelligence-generated videos, fraudulent QR codes and misleading posts that have flooded their platforms in the wake of the devastating Bhotekoshi river floods. In high-level talks held on Aug 28 with Nepal-based representatives of both companies, government officials sought faster and more proactive intervention against content that could mislead the public, obstruct relief operations and deepen the trauma of families whose loved ones were affected by the disaster. The meeting came after the Aug 26 flash floods triggered a torrent of misinformation across the digital landscape, ranging from recirculated archival clips presented as current footage to sophisticated synthetic media produced by AI tools. Officials responsible for regulating Nepal’s digital ecosystem said the situation had become serious enough to warrant direct coordination with platform providers rather than relying on conventional blocking measures. They argued that social media companies have both the technical capability and the moral responsibility to prevent their algorithms from amplifying harmful content in a time of national emergency. The request underscores the growing challenge that developing nations face in balancing open communication with the urgent need to contain viral falsehoods and digital scams during natural disasters. As rescue teams struggled to reach affected areas and relief supplies moved through difficult terrain, the authorities became increasingly alarmed that fake donation drives and fabricated updates were diverting attention and resources away from actual needs. The government’s response reflects a broader recognition that the online information ecosystem now plays as critical a role in humanitarian crises as physical supply chains, and that unregulated social media can turn a natural disaster into a secondary crisis of confusion and exploitation. With many Nepalis relying on Facebook, Instagram and TikTok for real-time updates, the officials made clear that the platforms must act as responsible partners rather than passive conduits for unverified and malicious material. The meeting marked one of the first instances in which Nepal has directly pressed global technology companies to modify their algorithmically driven content moderation practices during a disaster, setting a potential precedent for other South Asian countries facing similar challenges.

Bijay Kumar Roy, coordinator of the social media regulation and management task force formed by the government and a director at the Nepal Telecommunications Authority (NTA), explained that officials specifically asked the companies to remove links to objectionable and deceptive content as soon as they are flagged. He noted that internet service providers are only capable of blocking entire web domains, a blunt instrument that cannot surgically excise individual viral posts, videos or images. This technical limitation made direct coordination with Meta and TikTok essential. The government also urged both companies to make their algorithms proactive, so that graphic and false content is automatically blocked or downgraded the moment it is posted, rather than waiting for complaints from users or regulators. The current reactive model, in which platforms respond to takedown requests only after content has gone viral, has proven inadequate for a fast-moving disaster situation. Roy’s remarks highlight a deeper structural issue: content moderation systems are often designed for normal conditions, not for sudden surges of high-emotion material that can be weaponised by bad actors. During the first 48 hours after the Bhotekoshi floods, social media feeds were saturated with images of swollen rivers, destroyed roads and casualties, some of which were real and some of which were either old or entirely fabricated. The visual nature of the content made it especially dangerous, as viewers are inclined to trust images and videos more readily than text, even when those files have been manipulated with AI tools. The platforms’ existing community guidelines technically prohibit graphic violence, misinformation and fraudulent activity, but enforcement has been inconsistent and slow. By asking for algorithm-level changes, the Nepali government is effectively challenging the companies to treat disaster misinformation with the same seriousness as terrorism-related content or child exploitation material, which are removed almost automatically. Whether the companies will agree to such algorithmic changes remains unclear, but the officials said the discussions were constructive and that both Meta and TikTok acknowledged the urgency of the situation. Roy’s statement also suggested that the government is prepared to escalate its regulatory pressure if voluntary cooperation does not produce measurable improvements within a short period. In the meantime, the NTA is working to identify gaps in current enforcement and to establish a rapid-response mechanism that can flag problematic posts around the clock.

The scale of the misinformation problem became apparent almost immediately after the flash floods struck on Aug 26. Old videos of previous Himalayan floods and landslides were recirculated with captions claiming they had been recorded at the Bhotekoshi river, while AI-synthesised videos and images were presented as authentic footage of the disaster. Some of these synthetic creations showed roads and bridges collapsing in ways that never happened, buildings being swept away in impossible angles, and rescue scenes that had no basis in reality. While many of these creations may have been intended as crude attempts to attract views, the authorities fear that some were deliberately produced to create chaos, influence public perception, or monetise emotional reactions by driving traffic to certain pages and channels. A particularly concerning development was the appearance of fraudulent QR codes circulated on social media and messaging platforms, ostensibly to solicit donations for flood victims. These codes, when scanned, direct users to fake payment portals or personal accounts rather than legitimate relief organisations. Officials warned that such scams are designed to exploit public sympathy and divert money intended for humanitarian assistance into private hands. The fraudulent codes have been traced to posts that mimic the branding and messaging of established charities and even government agencies, making them difficult for ordinary users to identify as fake. Beyond the direct financial harm, these scams damage public trust in legitimate fundraising efforts, potentially reducing donations at a time when relief operations are heavily dependent on private generosity. The authorities have urged people to give only through verified official channels and to be suspicious of unsolicited payment links or QR codes shared on social media. They have also asked banks and digital payment providers to monitor for suspicious transactions linked to flood-related appeals. The combination of AI-generated disinformation, recirculated old footage, and donation fraud has created a complex information environment in which even well-intentioned users can inadvertently amplify harmful material. The government’s request to Meta and TikTok therefore goes beyond simple content removal; it calls for a coordinated response that includes source identification, account suspension, and public warnings about known falsehoods. Without such measures, each hour of delay allows the false narratives to harden and spread into private groups and messaging chains, where they are much harder to counteract.

Meta and TikTok have designated focal points in Nepal through which the authorities can flag problematic posts, and the companies have been cooperating with officials to remove material after receiving complaints. Roy confirmed, however, that the response is not always immediate, with platforms currently taking between 24 and 48 hours to act on reported posts. Such delays are extremely costly in a crisis, as misleading or distressing content can be shared hundreds of thousands of times in a few hours, moving beyond the reach of any single takedown request. By the time a post is removed, it may have already been downloaded, reposted, and subtly changed to evade detection. The government has therefore asked the companies to drastically shorten their response times and to implement emergency content-moderation protocols that can be activated during disasters. Officials have also specifically requested that platforms automatically blur graphic images and videos, particularly those showing victims and the aftermath of the disaster, before a human reviewer can assess whether they violate community standards. This approach is used in some contexts for disturbing content, but it is not standard practice for general content moderation. The request is driven by concerns that graphic images and videos shared without context or verification could deepen the trauma suffered by families of victims, who may encounter the remains of their loved ones circulating in their social media feeds. Mental health experts have long warned that exposure to such material can have severe psychological effects, including acute stress, post-traumatic stress disorder, and prolonged grief. In Nepal, where social media is deeply integrated into daily life and family networks, such exposure is nearly impossible to avoid once a video has gone viral. The government’s appeal for blurring and faster takedowns is thus not only about preventing misinformation but also about protecting the dignity and mental health of survivors and victims’ families. It also reflects a broader societal expectation that technology platforms should design their systems with greater sensitivity to the cultural and emotional contexts in which their content is consumed. The officials acknowledged that implementing automatic blurring across all content would be difficult, but insisted that emergency situations require exceptional measures. They proposed that Meta and TikTok develop a crisis-response framework that can be activated in Nepal and potentially other disaster-prone countries in the region.

To coordinate the government’s response, the authorities have formed a multi-agency team tasked with monitoring and regulating harmful online content throughout the crisis. The team includes representatives from the Ministry of Communication and Information (MoCIT), the NTA and the Ministry of Home Affairs, giving it a broad mandate that spans technical regulation, policy oversight and law enforcement. Within this structure, the NTA is responsible for technical coordination with social media platforms, including processing takedown requests and tracking the effectiveness of the platforms’ responses. The Nepal Police, meanwhile, is focusing on identifying individuals suspected of creating or circulating malicious content, and is preparing to initiate legal proceedings where necessary. The government-mobilised team consists of NTA director Roy, NTA deputy director Surya Prasad Lamichhane, assistant director Om Prakash Mahato, MoCIT engineer Kumar Yadav and inspector Arun Tamang from the home ministry. The composition reflects a deliberate effort to combine technical expertise with law-enforcement capacity. The police investigative component is particularly important, because many of the most damaging posts are not random acts of carelessness but organised attempts at fraud, disinformation and influence operation. Legal cases could be built under existing Nepali laws related to electronic transactions, cybercrime, and public order, though officials have not yet specified which charges may be applied. The multi-agency approach also allows for rapid sharing of intelligence across institutions; for example, when a fraudulent QR code is identified by the technical team, the police can immediately trace the associated accounts and payment networks. Similarly, when police identify a coordinated disinformation campaign, the NTA can move quickly to request account suspensions from Meta and TikTok. This level of coordination is relatively new for Nepal, which has historically struggled with institutional silos. The formation of the task force signals that the government considers social media regulation a top priority in disaster management, and the tools and relationships being tested during this crisis may well become permanent fixtures of Nepal’s regulatory infrastructure. At the same time, the team faces significant challenges, including limited staffing, the sheer volume of posts, and the technical sophistication required to detect AI-generated content. The authorities have not publicly disclosed how many posts have been flagged or removed, but the ongoing effort suggests that the problem remains acute.

In a press statement issued after the meeting, the communication ministry urged national and local media outlets and social media users to avoid sharing distressing photographs, unverified reports and fabricated material. The statement emphasised that information circulating during a crisis can have a significant impact on rescue operations, relief distribution and public morale. “We appeal to all national and local media outlets as well as social media users not to share unverified, fabricated, or AI-generated rumours that spread panic,” the ministry said. It also warned that graphic images and videos could deepen the trauma suffered by families of victims, particularly when such material is shared without context or verification. The authorities have urged people to verify information before forwarding or reposting it, especially material purporting to show the latest flood damage, casualties, rescue operations or requests for financial assistance. This plea is directed at ordinary users as much as at media organisations, because the real drivers of viral misinformation are often individuals who share content out of concern or solidarity without checking its authenticity. The ministry’s statement reflects a growing awareness in South Asia that misinformation is not only a problem of foreign trolls or malicious actors but also a community-level dynamic in which everyday sharing habits can contribute to the damage. In the days ahead, the government is expected to release additional guidance on how to identify AI-generated videos and verify official sources. Meanwhile, the pressure on Meta and TikTok remains high, and the outcome of these discussions will be closely watched by other governments in the region that face similar challenges during floods, earthquakes and other natural disasters. For Nepal, the Bhotekoshi floods have become more than a test of physical infrastructure; they have become a test of the country’s ability to govern the digital public square in an era of generative AI, viral deception and globalised content platforms. The response will determine not only how this disaster is remembered but also how prepared Nepal is for the next crisis.