The government of Jamaica was forced to deny a wave of fake videos that circulate on social networks, which show supposed disasters caused by the Hurricane Melissa.
The alert arises in the midst of the uncertainty caused by the onslaught of the cyclone, while videos, created through artificial intelligence (AI)add up to millions of reproductions and generate confusion among the local and global population.
Among the most viral videos is the disturbing scene of four sharks swimming in the pool of a Jamaican hotelsupposedly carried there by hurricane-related flooding. Another footage shows the Kingston airport completely devastated by the storm. However, none of these situations have actually occurred.
The country’s own Ministry of Education has highlighted that this is manufactured content that has spread through platforms such as x, TikTok and instagram.
Dana Morris DixonMinister of Education of Jamaica, warned about the danger of giving credence to these videos. “I am in many WhatsApp groups and I receive all these videos. Many of them are false. We ask you to please follow the official channels,” said the official in statements collected by Associated Press.
These examples are part of a growing trend: AI-generated videos have marked a new era in the spread of misinformation during natural disasters. Unlike traditional photographic montages, now the ease of creating hyper-realistic clips has moved the border between reality and fiction.
The problem is even more acute because fake videos created by AI are spread alongside genuine images recorded by residents and the media, which sows doubts and confuses users of social networks.
Expert recommendations for identifying manufactured content include searching for logos or watermarks from tools such as sorathe video-to-text generator developed by OpenAI.
However, it is becoming easier to eliminate these traces using editing programs, so it is suggested to observe carefully if there are blurs, pixelated areas or inexplicable color changes where there should originally have been a watermark.
Fuzzy details, objects jumbled together in strange shapes, or illegible text on signs are common evidence of fabricated videos. For example, although the video of sharks in the pool may appear authentic at first glance, closer analysis reveals anomalous shapes in some animals, which reveals digital manipulation.
Sofía Rubinson, senior editor at NewsGuardan organization dedicated to analyzing online misinformation, explained: “Now, with the rise of tools as accessible and powerful as Sorait is even simpler than unscrupulous actors create and distribute highly compelling synthetic videos”.
“Before, it was possible to detect forgeries by unnatural movement, distorted text or details such as missing fingers. But as technologies improve, many of those defects disappearand for the average viewer it is increasingly difficult to identify AI-generated content,” Rubinson added.
The rapid expansion of this content does not necessarily respond to political agendas, but rather to economic incentives and digital notoriety. The AI specialist Henry Ajder pointed out that most of the fake videos about Hurricane Melissa respond to the logic of traditional clickbait: they seek to increase the engagement and, therefore, the financial rewards or the growth of followers.
On platforms like X, users can get paid based on the level of engagement on their posts. Likewise, in YouTubecreators earn income by displaying ads on their videos. “A video that reaches millions of views can earn the author several thousand dollars, which is very attractive in relation to the effort invested“Ajder explained.
Additionally, many social media accounts use these strategies to increase their follower base and then promote products or services. It is therefore essential to verify the history of whoever publishes this type of content: if it is an account associated with sensational headlines or of dubious veracity, it is advisable to distrust the authenticity of the material.
Not all creators seek to hide their identity. “Some simply want to experiment with artificial intelligence to attract attention”he mentioned Ajder.
The authorities, together with meteorological experts, reiterated the need to use only official channels to obtain information about Hurricane Melissa and its consequences. “Do not depend on strangers on the internet for information. The Jamaican government and the National Hurricane Center are updating the news on their official sites and profiles,” emphasized Dana Morris Dixon in dialogue with AP.
In the case of the shark video, although a version on Instagram shows the seal of a TikTok profile —Yulian_Studios—, said account defines itself as “content creator with AI visual effects in the Dominican Republic.”
While the shark clip no longer appears there, other AI videos do exist, such as one showing a man clinging to a palm tree during supposed hurricane-force winds in Jamaica.
Platforms are also taking action. In X, for example, a version of the pool video includes a clarifying community note: “This video and the voice used They were created using artificial intelligenceit is not a real record of the Hurricane Melissa in Jamaica”.
Experts in digital verification, like the journalism site Poynterrecommend questioning videos that look “exaggerated, unrealistic or far from what is expected” and analyzing both the image and the audio.
Although until recently fake virtual announcers displayed unnatural voices, technology has advanced to the point of achieve perfectly synchronized and convincing sounds.
