CBS LA's Kristine Lazar Misses 9 of 12 Deepfakes, Then Lifts Score to 8
Updated
Updated · CBS New York · Jul 18
CBS LA's Kristine Lazar Misses 9 of 12 Deepfakes, Then Lifts Score to 8
3 articles · Updated · CBS New York · Jul 18
Summary
Kristine Lazar correctly identified just 3 of 12 images and videos in an AI deepfake quiz, a 25% hit rate that mirrored a survey finding Americans perform little better than chance.
Veriff said older visual giveaways such as warped fingers or unnatural eyes have largely vanished as AI image quality improved, leaving viewers to spot subtler clues in facial texture, patterns and motion.
After guidance from Veriff product director Raul Liive, Lazar retook the quiz and scored 8 out of 12, catching details such as mismatched earrings that she initially missed.
Liive said even specialists cannot reliably spot every fake without extra tools, reinforcing the report's takeaway that visual judgment alone is no longer enough to verify images or video.
With human eyes now failing to spot fakes, can advanced AI detectors truly win the escalating deepfake arms race?
A $25 million heist used a fake video call. What non-technical defenses are crucial to stop the next corporate deepfake scam?
How does society rebuild trust when any video or audio can be convincingly faked in mere seconds?
2026 Deepfakes Crisis: Veriff Report Reveals 3 in 4 People Can’t Spot AI-Generated Content—Impacts, Risks, and Solutions
Overview
Veriff's 2026 Deepfakes Report, created with Kantar, reveals that most people cannot detect AI-generated content and often misunderstand its nature. By testing 3,000 participants with real and fake visuals, the study shows the public is highly vulnerable to deepfakes. This widespread inability to spot manipulated media creates a risk of major societal disruption, as people may trust false information. The report highlights an urgent need to address deepfakes, since public misidentification can erode trust and make society more susceptible to deception. These findings stress the importance of improving awareness and defenses against AI-generated content.