Google has added an important new feature to Gemini, giving the AI chatbot the ability to check whether an image was created with artificial intelligence. The update arrives shortly after Google launched Gemini 3, but this new detection tool is separate from the model’s upgraded generative abilities.
The new option is based on Google’s SynthID Detector, a system the company introduced earlier this year. SynthID places an invisible digital watermark inside images, videos, audio, and even text created with AI. Unlike traditional watermarks, these markers are hidden within the file’s data and cannot be seen inside the image itself.
This watermarking system is already built into several Google tools, including Gemini. Now, users can bring the detector directly into their chats. By tagging @SynthID and uploading an image, Gemini can scan the file and report whether the hidden watermark appears.
How the Detection Works
Gemini checks for traces of SynthID’s watermark and shows how much of the mark is present. If only part of an image was generated or edited with AI, the detector can pick up those sections as well. This makes it useful for spotting AI-edited content, not just fully AI-generated work.
The system is designed to help identify digital content that may look convincingly real, which is becoming more important as AI tools become easier to use and widely available.
The main limitation is adoption. SynthID works only on content generated by tools that support Google’s watermarking system. Many other AI platforms still do not include these markers. Google says it is working to bring more partners on board so that detection becomes more reliable across the industry.
