As dozens of videos are being shared online following the deadly floods in Nepal, questions are emerging over the authenticity of some of the footage. A dramatic video purportedly showing the Nepal floods has been identified as AI-generated, according to the BBC.
The publication has outlined several ways viewers can check whether a video has been created using artificial intelligence.
Check The Video Length
One of the first red flags can be the duration of the footage. The BBC analysis noted that the Nepal flood video shared online was only around 10 seconds long. While video-generation tools are rapidly evolving, some AI models, particularly free versions, have historically imposed limits on the length of clips they can generate.
ALSO READ: Nepal PM, Cabinet Ministers To Forego One-Month Salary For Flood Relief Funds
"Current AI models, at least the free ones, are limited to generating up to 15 seconds of footage," the publication said.
However, video length alone is not proof that footage is AI-generated.
Use AI Detection Tools
If a video looks unusually polished or simply seems "too good to be true", viewers can download it and analyse it using available AI-detection tools.
Technology companies have also started introducing systems designed to identify content generated using their own AI models. In this case, analysis using Google's SynthID indicated that the footage had been generated using Gemini.
Such tools can provide useful clues, although their results should not necessarily be treated as conclusive proof on their own.
Slow Down And Look Closely
Another way to spot AI-generated footage is to slow the video down and examine individual frames.
The BBC found several inconsistencies in the Nepal flood footage. Red containers seen in one part of the video appear to disappear, while a log seems to appear suddenly without any clear explanation.
Objects, people and backgrounds can sometimes change shape, disappear or appear unexpectedly in AI-generated footage.
Ask: Does It Make Sense In Real Life?
Viewers should also look for things that do not make physical or logical sense. AI-generated videos can struggle with consistency, particularly when depicting fast-moving scenes involving water, vehicles, people or multiple objects. Unnatural movement, changing objects, distorted details and impossible interactions can all be warning signs.
Notably, AI-generated disaster videos are appearing increasingly frequently on social media. Some may be created to attract engagement, while others can be deliberately produced to spread misinformation.
ALSO READ: Nepal On Alert: China Issues Warning As Overflowing Lake In Tibet Raises Fresh Flood Risk
The problem is compounded when genuine users share such videos believing them to be real.
In the case of Nepal, there is already substantial genuine footage showing people affected by the floods. When AI-generated material is circulated alongside authentic images and videos without verification, it becomes increasingly difficult for users to distinguish fact from fiction.
The key process is simple: check the source, examine the video closely, use available detection tools and verify the footage against credible reports before sharing it.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.