How to check if the image is AI-generated? (Jimmy Kimmel case. Part 2)
The viral image of Jimmy Kimmel with a fake ‘final monologue’ sign was 70% AI-generated. The heatmap gave away the digital seams.
Jimmy Kimmel is not leaving his show. Part 2
Here’s the first part of our investigation on the Facebook story of Jimmy Kimmel’s quit from his show due to the conflict with Donald Trump. We already checked the AI-generated text (99% fake) and now it’s time for the image. We used our AI image detector, and its heatmap lit up like a Christmas tree, scoring the photo 70% AI-generated. This is how you kill a viral lie twice: first the story, then the ‘proof.’
All our AI detection models are brutally specific. Here again, image detector didn't just say ‘fake’, but highlighted exactly where the AI struggled: the cardboard sign with its text and the upper half of Kimmel's body and suit. This tells us exactly which parts were heavily altered to create this piece of AI-generated engagement bait.
What gives out the fake?
Here’s the breakdown of the digital tells our detector flagged:
- the skin looks unnatural,
- all the textures look unnatural as well – look at Jimmy’s suit and the cardboard,
- people’s faces on the background are blurred and do not really have realistic facial features or expressions,
- the text on the sign looks printed, all the letters have the same ‘font’, though it’s implied that this text is written by hand.
Source: Snopes fact-check report
FAQ
Is the photo real?
No. Our AI image detector scored it 70% AI-generated. The heatmap specifically highlighted the sign and Kimmel's upper body as the most synthetic parts, confirming it's a digital fabrication.
How to check AI-generated images?
Easily. With isFake AI image detector. Upload the pic and get the result in seconds. Additionally, our heatmap highlights the most AI altered areas of the image.
What tool can detect Midjourney images?
isFake AI image detector is built to defend you against AI fakes. The detector performs 98% detection accuracy and not only gives the verdict ‘AI’ or ‘Human’, but provides the confidence score(%) and the heatmap with the most suspicious areas highlighted.
