(Credit: Google/YouTube)
It has become increasingly difficult to tell the difference between real videos and AI-generated ones. Google's Gemini has a new tool to help you sort fact from fiction, but there's a big limitation.
Gemini will now be able to identify videos generated using Google's own AI models, the company said in a blog post. Just upload the video you're doubtful about and ask simple questions, like "Was this video generated using AI?"
To answer your question, Gemini will look for Google's proprietary SynthID watermark, which it adds to all media generated using its AI tools. Gemini's response will be a simple yes or no, combined with details about which part of the audio or video had the SynthID watermark.
At launch, the feature works for files up to 100MB and 90 seconds long. A similar feature for AI image detection was rolled out last month.
Additionally, Gemini also seems to have added a handy new markup tool for its image edits. After uploading a photo, you can tap on it to add markings or text to indicate the changes you want. This is useful when you want to add or remove an element from a specific part of the image.
(Credit: PCMag)The feature's exact use cases are unclear. It was spotted by Android Authority, and Google hasn't officially announced it. Just a note of caution here: after adding your markups, you need to click Done and ask the chatbot to remove them from the final output.
Last month, Google rolled out its most capable Gemini 3 model, which outperformed OpenAI's GPT 5.1 in several AI benchmarks.
Following Gemini's success with image generator Nano Banana and a leap in performance with Gemini 3, it was reported that Sam Altman had issued a "code red" alert to his employees at OpenAI, asking them to work on improving ChatGPT's everyday features. This month, the company released GPT 5.2 and a new image generation model.