Seeing is no longer believing.
Fake photos, voices and videos made by AI now look real.
People who can spot fakes will pull further ahead of people who can’t. 76% of Americans say it is important to know whether content was made by AI, but 53% are not confident they could tell.1 When U.S. adults were tested on 16 real and fake visuals, their accuracy was no better than guessing.2
53% of Americans are not confident they could tell whether content was made by AI or a person (Pew Research Center, 2025).
More numbers
of American adults are familiar with the term “deepfake.”2
of U.S. respondents picked the deepfake as the real video in one side-by-side test.2
of Americans say they no longer know what is real online at all.3
“Many don’t trust their own ability to spot AI-generated content.”
Pew Research Center, 2025
The other side
Detection tools and content labels are improving, and most people still see far more real media than fake.
accuracy for a commercial deepfake detector tested on 300 fake videos. TEM Journal, 2025 ↗
accuracy for a well-tuned detection model on a standard benchmark. PLOS One, 2025 ↗
Why Anthony still holds this view
Detectors work in labs, accuracy drops on compressed real-world video, and people don’t run detectors on their feeds.
For marketers
Label your AI-made content and show proof of what is real: behind-the-scenes footage, verified accounts, named people.
For you
Before sharing a shocking image or clip, look for a second source.