Presented by Veriff
Americans struggle to distinguish between real and AI-generated content, posing significant challenges not only for media literacy but also for online identity verification by businesses.
A recent study reveals that while awareness of deepfakes is widespread, the ability to differentiate them from real content is scarcely better than random guessing. Conducted by Veriff and Kantar in 2026, the survey involved 3,000 participants from the United States, the United Kingdom, and Brazil, with Americans scoring only 0.07 on a scale where zero means random guessing.
This inability to discern authentic visual content means users can’t reliably verify identities online. Consequently, people using digital services may struggle to identify if the person they are interacting with is genuine.
This challenge has significant implications for digital businesses that rely on image and video-based identity verification, affecting processes such as customer bank onboarding, account recovery, marketplace seller checks, high-value e-commerce transactions, social platform authentication, and enterprise access control.
In the U.S., these issues are already causing substantial losses, with synthetic identity fraud contributing to billions in annual losses, and the technology to create convincing fakes becoming increasingly accessible.
The report highlights a small, high-risk group: about 7% of users who fail to detect deepfakes effectively but remain confident in their abilities, seldom verifying what they see. Although small in percentage, this group represents millions of accounts vulnerable to exploitation.
When users can’t distinguish real from synthetic identities, systems relying on visual verification are fundamentally exposed. Identity verification must be integrated as a core component of digital infrastructure rather than merely a compliance requirement.
“As AI-generated content becomes indistinguishable from reality, the human eye is no longer a dependable defense,” states Ira Bondar-Mucci, fraud platform lead at Veriff. “U.S. businesses and policymakers must address this awareness gap promptly while investing in automated verification technologies that can detect what humans cannot.”
The U.S. deepfake awareness gap is wider than expected
Though the U.S. is a leader in generative AI development, American consumers are the least familiar with deepfakes among the surveyed countries. Only 63% of U.S. adults are familiar with the term, compared to 74% in the UK and 67% in Brazil.
“A paradox exists,” Bondar-Mucci explains. “Despite being the global hub of AI development, Americans are less aware of its dangerous byproducts. Historically, the conversation has focused more on data privacy than content authenticity, leading to a higher baseline trust in digital content. Low awareness doesn’t mitigate risk; it increases it. Without knowledge of deepfakes, individuals are less likely to scrutinize and verify what they encounter.”
Human deepfake detection is barely better than a coin flip
Consumer ability to distinguish real from fake content often resembles random guessing, especially for video content. Fake videos are frequently perceived as authentic, and real ones are sometimes deemed fake. Even when comparing side-by-side, judgments are nearly split, showing that visual inspection alone is no longer a reliable means of verification.
Overconfidence in deepfake detection creates a dangerous vulnerability
Approximately half of U.S. respondents express confidence in spotting deepfakes, yet their performance doesn’t support this confidence, making self-assessment ineffective.
Among these respondents is a small but risky group: around 7% who are poor at detecting deepfakes yet remain overconfident and rarely verify suspicious content.
“This gap between confidence and competence creates a false security sense that fraudsters can exploit,” Bondar-Mucci warns. “When people believe they can’t be fooled, they stop being vigilant, making them vulnerable to synthetic identities used in fraud or fabricated videos designed to manipulate trust.”
For businesses, this signifies that relying on manual review processes or customer self-assessment exposes them to vulnerabilities. Human judgment is increasingly unreliable, necessitating that verification processes be automated and technology-driven, not reliant on the end user’s ability to distinguish real from fake.
Americans are worried about deepfakes but trust platforms to handle them
Americans are highly concerned about deepfakes, with 79% expressing concern about personal fraud and impersonation.
Unlike other markets, Americans are more inclined to trust social media platforms and digital services to manage AI-generated content, potentially reducing individual vigilance just as the threat is growing.
“Synthetic identities are being used to open fraudulent accounts and authorize transactions, and deepfake videos are circumventing basic verification checks,” Bondar-Mucci notes. “The urgency lies in the combination of high concern and high platform trust. The gap between perceived and actual protection is where fraud prospers.”
The business case for automated identity verification has never been stronger
The disparity between what Americans believe they can detect and what they actually can is not a knowledge issue that awareness campaigns alone can resolve, but a flaw in any system that relies on human judgment for identity verification.
The solution is not to remove humans entirely from the verification process, but to cease depending on them for tasks that human perception can no longer reliably perform. Organizations persisting with manual reviews or customer self-attestation are embedding vulnerability into their operations.
The alternative is automated, AI-driven identity verification at the point of interaction, capable of detecting synthetic media before human intervention is needed, and independent of the user’s ability to discern authenticity.
“Seeing is no longer believing,” Bondar-Mucci concludes. “Companies that integrate verification infrastructure based on this reality, rather than assuming otherwise, will be best positioned to maintain customer trust as the synthetic media landscape evolves.”
Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.
