AI Videos Falsely Claim Kamchatka Snowstorm

Impressive videos circulating on social media claim to show the aftermath of a historic snowstorm in Kamchatka, Russia. The footage features buildings buried in snow and children sliding down frozen walls.

However, analysis reveals the footage is entirely synthetic. Independent verification tools confirm the videos are AI-generated fabrications, not documentation of the region's actual weather events.

Three specific videos gained traction on X (formerly Twitter) on Monday, January 19. The timing coincided with the heaviest snowfall to hit the Kamchatka peninsula in 60 years.

The first video shows a building completely covered in snow, with children climbing out a window and sliding down the massive drift. The Russian audio translates to: "Nossa, parece que cai direto do telhado. Este é o quarto andar, é assustador. Mas é divertido" (Wow, it looks like it fell right off the roof. This is the fourth floor, it's scary. But it's fun).

A second clip displays half of an edifice buried under snow, while a third compilation shows streets lined with towering snow walls where people walk between them. Captions in English claim the "winter arrived in Kamchatka" and that "life continues with impressive calm" amid the frozen landscape.

Forensic analysis of the videos revealed consistent synthetic markers. HiveModeration, a tool designed to detect AI-manipulated media, analyzed all three clips.

The results were definitive:

• Video 1: 99.9% probability of AI-generated images and audio
• Video 2: 99.9% probability of AI generation
• Video 3: 99.9% probability of AI generation

A visual anomaly in the first video provided immediate evidence of fabrication: a child sliding down the snowbank leaves no footprints or trail in the snow.

"Feito com IA do Google (vídeo) – Synth ID identificado em todo ou parte do conteúdo carregado"

The third video underwent additional scrutiny using Google's SynthID Detector. This tool identified a digital watermark embedded directly into the video frames—a technique imperceptible to human viewers but traceable by Google's systems. The analysis confirmed the presence of this watermark across nearly the entire video.

Reverse image searches using Google Lens traced the videos to their actual sources, none of which were in Russia.

The first video, featuring the "snow slide," was originally posted on Instagram on January 18 by a content creator specializing in AI simulations. The hashtags in the caption explicitly indicated the synthetic nature of the content.

The second video appeared on TikTok on January 16, posted by an account dedicated to AI-generated content.

The third video, which SynthID identified as Google AI content, was published by a graphic designer based in Ankara, Turkey on January 12. The original caption explained the footage was a simulation of how the Turkish city would look covered in snow.

The misinformation campaign exploited a genuine weather event. On the same day the videos circulated, Kamchatka experienced its most severe snowstorm in six decades.

The actual storm created massive snowdrifts, blocked building entrances, and buried vehicles. This real meteorological event provided the perfect backdrop for the fabricated videos to gain credibility.

By attaching the footage to a real news event occurring at that exact moment, the creators ensured the content would be perceived as authentic documentation of the historic weather.

This incident demonstrates the increasing sophistication of AI-generated misinformation and its ability to exploit real-world events. The videos were created and distributed during the actual storm, making them appear timely and relevant.

Verification tools like HiveModeration and SynthID provide essential capabilities for detecting synthetic content, but the speed at which misinformation spreads often outpaces fact-checking efforts.

As AI video generation technology advances, distinguishing between authentic footage and synthetic creations will require both technical tools and increased media literacy among consumers.