Social media posts on flooding in Thailand share AI-generated video

A fabricated clip has circulated in social media posts falsely claiming it showed flooding in Thailand after rainfall and strong winds from Tropical Storm Bualoi devastated several Southeast Asian countries in late September. The clip contains several visual inconsistencies and is also watermarked with the icon of an AI generative model from Google 

"In Thailand, many people went missing. Now the planes are searching for them downstream," reads the Khmer-language description of a Facebook reel published on October 3, 2025.

The video shows houses submerged by floodwaters while dozens of people await rescue on crowded rooftops.

Image
Screenshot of the false post taken on October 8, 2025, with a red X added by AFP

Severe Tropical Storm Bualoi made landfall in Vietnam and damaged more than 200,000 homes, killing at least 50 people across the country (archived link).

Floods triggered by Bualoi and heavy rain have also affected more than 33,000 families in Cambodia and forced the evacuation of thousands of households.

Thailand was not hit by the storm, but authorities said on September 27 the death toll from floods triggered by heavy monsoon rains in the kingdom rose to seven as relief efforts continued (archived link). 

More than 260,000 people across several provinces along the Chao Phraya river were impacted by flooding, the Department of Disaster Prevention and Mitigation said in a statement.

AFP has debunked a number of posts in which users in both Thailand and Cambodia spread false claims with misrepresented visuals, suggesting the other country was repeatedly struck by catastrophes following deadly border clashes in July.

The false video has been shared alongside similar claims on Facebook and Instagram in Cambodia, while Facebook users in Vietnam shared a screenshot from the video and claimed it showed flooding in Thanh Hóa province.

But it was in fact generated using AI.

A reverse image search on Google using keyframes revealed it was labelled as "Made with Google AI".

Google's SynthID Detector -- a tool launched in May 2025 to identify AI-generated content -- detected digital watermarks that indicated it was created with the tech giant's generative model (archived link).

Image
Screenshot from the SynthID Detector tool, with highlights added by AFP

A star icon on the lower right corner of the clip indicates it was created with Gemini, Google's AI assistant tool (archived link).

An analysis of the footage also found visual inconsistencies indicative of AI-generated content, including distorted faces, disproportionate limbs, and a man in a white T-shirt moving unnaturally while other elements in the frame remain still.

Image
Screenshot of the falsely shared image, with visual inconsistencies magnified and highlighted by AFP

AFP has previously debunked other false claims spreading AI-generated visuals.

Is there content that you would like AFP to fact-check? Get in touch.

Contact us