Google Gemini gets an artificial intelligence video detector

It’s getting harder and harder to tell if the video you’re receiving is a real recording or computer-generated content. However, you can now load it and check the result. The process is very straightforward. Just upload the video to Gemini and ask a simple question; For example, ask if this video was produced using Google’s artificial intelligence.

Gemini starts the verification process with a technology called SynthID. SynthID is Google’s proprietary digital watermarking technology that embeds signals in content generated by artificial intelligence. These signals cannot be detected by humans, but they are easily detected by software. The tool performs a thorough check and analyzes the entire file to determine whether AI was used for the background music, the video images themselves, or both.

If you plan to use this tool continuously, you should be aware of some practical limitations. Currently, the files you upload should not exceed 100MB in size and their duration should not exceed 90 seconds. These limitations mean that it is not possible to check full movies, but it will be enough to check short clips and social media content.

This new feature is an extension of the tool that Google had previously launched for images. The company has been continuously advancing its SynthID technology for a considerable period of time, aiming to provide structured transparency to the content generated by its tools. Since the introduction of this technology in 2023, Google claims to have watermarked more than 20 billion pieces of content produced by artificial intelligence. Such a high volume of flagged content means that if an AI-generated image originates from a Google-owned model, Gemini’s system can identify it almost immediately. This new feature extends the same level of scrutiny and rigor to moving images and audio content.

However, there is one very important limitation that should be noted. This tool is strictly limited to content that has been produced or edited using Google’s own internal tools. If an image or video is created using an artificial intelligence model not affiliated with Google, Gemini will not be able to provide information about it. As a result, this tool is practically only used to create transparency within Google’s exclusive ecosystem.

Google prefers users to rely on Gemini for such reviews and not need to use third-party tools to analyze images and videos. Although this company has made it easier to identify the content produced by itself, the lack of support for external models makes this feature not considered as a comprehensive and universal tool for identifying artificial intelligence content.