While some AI-generated videos are obvious, it's more difficult to decipher whether some videos are real or not. Instead of assuming or hoping that someone in the comments knows the answer, you can now identify videos generated by AI using Google Gemini.
The dangers of AI-generated video
At first, AI videos may seem harmless. Often they are just cute animals or atmospheric videos. But now, real creators are masquerading as legitimate videos, offering helpful advice and encouraging viewers to download files, try troubleshooting solutions and hacks.
The problem is that the advice, files, solutions, and hacks are either completely false or malicious. An entire ghost network of AI-generated videos on YouTube lured thousands of users into downloading malicious files. It doesn't even have to be technology related. These videos may provide health advice. You might think it was written by a real expert, but it's all completely fake.
The problem is sharing. Watch videos that are obviously real and share them on social media. Instead, it is generated by AI. Sharing fake content can lead to losing your followers and even your job.
Google Gemini wants to help you tell the difference between reality and AI with new AI video detection capabilities. Currently, ChatGPT also does not offer this functionality.
How Gemini detects AI videos
Currently, this feature is limited to AI videos and images created with Google's AI tools. Everything Google's AI creates includes an invisible watermark that you can't remove. The SynthID watermark is present within all audio, image, and video AI created by Google.
The Gemini app scans the SynthID watermark to determine if AI is used. Even videos and images that have only been edited with AI still contain watermarks.
Check out videos generated by Google AI
First, download the video in question. You may need a third-party tool or site to do this. If you want to download YouTube Shorts, try these tips.
Log in to your Gemini app (mobile or web). next, + Click the button in the chat box to upload your video. The app accepts up to 90 seconds or 100MB. Once uploaded, ask Gemini if the video was created using AI.
It usually takes less than 30 seconds to get a response. What I like most is that Gemini not only tells you if it's AI, it also tells you why the video is AI. If your video has a SynthID watermark, Gemini will notify you that it has a watermark and will also note other signs that the content is an AI-generated video.
To test it, I downloaded an AI clip from Gemini's Veo site and asked Gemini if the video was created with AI. The AI provided a positive answer within seconds.
Testing AI on every video
Although we can't guarantee it, Gemini can also test the AI content of any video. In my testing, it was very good at detecting AI-generated videos created with other tools.
You can download clips of videos or paste links to videos. Sometimes Gemini was very confused and could not quite understand what the link was. For example, Gemini thought the Gingerbread Village video on YouTube was an AirHead video. I had to ask the question again to get the correct result.
I tried some videos from YouTube as a test. To avoid confusion, I used the following prompt:What is this video about YouTube links? Does it use AI?”.
We were confident that Video A was generated by AI, and Gemini was right.
I tried another video that I know for sure doesn't use AI. It was created before AI, and its creator is famous for not using AI even today. Gemini had no problem explaining why it wasn't an AI-generated video.
Test images for AI too
It's not just limited to videos. Gemini will also test images for AI. Again, it works best with images created with Google's AI tools, such as Nano Banana.
The process works the same way. Upload an image and ask Gemini if the image uses AI.
For both images and videos, Gemini will often indicate whether it is fully or partially AI. For example, a creator may have used AI to create a base image or background for a video, then manually edited parts of it using other tools to get the final result. This is a hybrid model.
AI is not fully guaranteed.
Like any AI detection tool, Google Gemini isn't perfect. It makes mistakes. In my testing alone, Gemini misunderstood the YouTube videos I was asking questions about.
However, it does a better job of detecting AI content than most other AI platforms. Additionally, this is a new feature from Google, so it's likely to further improve pointing out AI-generated videos and images across all platforms.
Other platforms like TikTok are starting to offer ways to detect and reduce AI-generated content for people who are tired of seeing it. You can also use other AI detection tools to determine what is real and what is fake.
