Associated Incidents
NEEDHAM, Mass. ---
A deepfake image of President Donald Trump that received millions of views on social media is the latest large-scale example of how generative artificial intelligence can be used to create falsehoods that grab and divert public attention.
Multiple reports indicate the photo was originally posted on X, the platform formerly known as Twitter, by a self-described Democratic strategist. It appears to show the president using a walker to move down a hallway, but a close inspection reveals several signs that it is fake.
Some possible signs of an AI image or video:
In lieu of an easy identifier, like a watermark, here are a few things to look for:
- Human details: AI frequently struggles with details, including fine hair strands or realistic hand anatomy.
- Backgrounds: Unnatural warping, inconsistent shadows, unusual textures, spotty blurring or strange patterns could be clues.
- Objects: Look for unnatural or asymmetrical shapes, incorrect reflectivity and mismatched sizes.
You can ask an AI for help:
Perhaps it's a bit ironic, but if you're not sure about an image, you can upload a photo or a link to an AI to get a jump start on your analysis.
We asked ChatGPT and Gemini to inspect the fake image of Trump. Both identified inconsistent background details, distortions, unnatural focus and middled textures as signs of likely manipulation.
AI Image Example
Gemini, a Google product, also impressed us by automatically identifying a SynthID digital watermark and matching the image to preexisting coverage from fact-checkers who debunked the image.
In this case, the SynthID watermark provides the definitive conclusion that the image is a fake.