Apparently he was using Nano (by Google) to create those fake pics. Nano is one of the best AI-image generators, with now very realistic features that can deceive people. However, for those who don't know, Gemini uses a technology called Synth-ID when creating images. This technology was introduced to fight disinformation (again, because Nano generated photos are now highly-realistic) and it's embedded inside the generated picture, meaning that can't be deleted by cropping, rotating, brightness/contrast changes, making screenshots etc... A 100% working method to see if a photo was created using Gemini/Nano is: 1. Go to the Gemini chat 2. Upload the photo and prompt Gemini with the words: "Check Synth-ID". 3. Gemini will tell you if the photo contains Synth-ID, meaning that it was created by it. Note that it won't work if a photo was created by ChatGPT or different models from Nano.
One of the photos he was using (and that looks real at first glance) was flagged by Gemini even though I took a screenshot from a news article:
