Unusually intense snowfall in Russia’s far eastern Kamchatka Peninsula paralysed the region and buried cars and buildings, but clips circulating online purportedly showing huge snowdrifts are in fact AI-generated. The clips, which contain visual errors often found in synthetically created content, were traced to images shared on Threads that the poster said were made using AI chatbot Grok.
“Snow apocalypse in Kamchatka,” reads part of the lengthy Malay-language caption of a Facebook video shared on January 20, 2026.
The video shows a montage of clips of snowdrifts several storeys high, some appearing to completely engulf blocks of flats.
Screenshot of the false post captured on January 27, 2026, with a red X added by AFP
Similar Facebook and TikTok posts also shared purported footage of the snowfall. It also spread in other language including English, French, Greek, Dutch and Romanian.
It circulated as Kamchatka recorded its heaviest snowfall since the 1970s in the first weeks of January 2026, prompting authorities to declare an emergency situation with schools closed and businesses forced to switch to remote work (archived here and here).
Snow falling from roofs killed two people. Images shared by the Russian authorities showed huge snow piles reaching up to the second storey of buildings and people digging their way through roads as snow blanketed cars on either side.
HandoutRUSSIAN EMERGENCY SITUATIONS MINISTRY
(Handout / RUSSIAN EMERGENCY SITUATIONS MINISTRY)
The circulating clips, however, were generated with the help of AI.
An analysis of the four clips comprising the falsely shared video using the Hive Moderation detection tool found they were “likely to contain AI-generated or deepfake content” (archived link).
Screenshot of the Hive Moderation detection results
Visual inconsistencies also suggest the footage is not authentic.
The snow appears unusually smooth and symmetrical, or breaks off abruptly in vertical lines, which appears unrealistic. Architectural details on windows and balconies also show inconsistencies and distortions typical of AI-generated content.
Screenshot of the false post with the visual anomalies highlighted in red by AFP
A reverse image search led to stills of the falsely shared clips shared in a Russian Threads post shared on January 17 by a user called “ibotoved“, who captioned them “Kamchatka today”.
The falsely shared clips appear to be animated versions of the images.
The account owner later said in the comments section that the images were AI-generated using Grok — an AI assistant developed by xAI — after several users questioned their authenticity.
On January 20, ibotoved posted again on Threads, saying the images he shared were fake. He described the images as a “prank” and added that “millions of people believed the fake photos.”
He attached a video that alternated between ordinary winter scenes and exaggerated images. One of the Russian-language captions read: “Don’t believe everything you see.”
AFP contacted the user on January 20, but had not received a response by the time of publication.