Deepfakes Fuel Growing Crisis of Trust in Digital Media

The rising tide of synthetic media in politics has taken an unexpected turn, challenging both experts’ predictions and public perception. While deepfakes were initially viewed primarily as weapons of disinformation that could undermine democratic processes, their actual deployment in political arenas has proven more nuanced and multifaceted.

The technology—which originated with a Reddit user called “deepfake” in 2017—has since evolved into a complex phenomenon with competing definitions. American and British linguistic traditions reflect subtle but significant differences in how deepfakes are characterized. Merriam-Webster defines them as altered media misrepresenting someone’s words or actions, while the Oxford English Dictionary adds an explicit moral dimension, highlighting their potential for malicious use.

These definitional differences extend to legislative approaches as well. The United States has pursued narrow definitions in proposed legislation like the TAKE IT DOWN Act and DEEPFAKE Act, focusing on impersonation and malicious intent. By contrast, China has adopted a broader approach, regulating virtually all AI-generated media regardless of intent. The European Union has positioned itself somewhere between these extremes, with the AI Act defining deepfakes as content that “would falsely appear to a person to be authentic or truthful.”

The 2024 global election cycle has showcased the diverse applications of deepfake technology. In India’s national elections, synthetic media served both disruptive and constructive purposes—creating false scandals involving Bollywood celebrities while also enabling politicians to deliver campaign messages in multiple languages to reach diverse constituencies. Similarly, the U.S. presidential race saw deepfakes deployed both as tools for defamation and as means for candidates to amplify their messaging.

Wartime applications have proven equally varied. Early in Russia’s invasion of Ukraine, a synthetic video showing President Volodymyr Zelensky announcing Ukraine’s surrender circulated as propaganda. Conversely, in conflict zones like Gaza, activists have used AI-generated content to circumvent content moderation policies that might otherwise block graphic depictions of war.

The theoretical concerns about deepfakes center on what researchers call the “realism heuristic”—the cognitive shortcut that leads people to believe what they see. The fear has been that visually convincing synthetic media would exploit this tendency, creating seemingly irrefutable evidence of events that never occurred. However, recent empirical studies suggest this might be an oversimplification. Deepfakes don’t necessarily prove more effective at misleading audiences than traditional forms of disinformation.

More concerning may be the systemic effects of deepfakes on information ecosystems. As philosopher Don Fallis notes, they represent an “epistemic threat” by undermining the very foundations of how we verify information. When visual evidence—traditionally one of the most trusted forms of proof—becomes questionable, our collective ability to gain new knowledge about the world is compromised.

Research has documented a troubling “spillover effect” where awareness of deepfakes decreases trust not only in manipulated content but in genuine information as well. Both laboratory experiments and field studies indicate that exposure to synthetic media correlates with lower levels of trust and perceived news credibility across the board.

This erosion of trust creates fertile ground for increased polarization. When “seeing is no longer believing,” deepfakes become convenient excuses for selectively accepting only information that aligns with pre-existing beliefs while dismissing contradictory evidence as artificial or manipulated.

The challenges posed by deepfakes are compounded by pre-existing vulnerabilities in media landscapes. Trust in traditional information sources has been declining for decades. The proliferation of synthetic media accelerates this trend while simultaneously revealing how fragile our information verification systems have become.

Paradoxically, the deepfake phenomenon may contain the seeds of its own solution. By exposing the limitations of our current approach to information verification, synthetic media could catalyze renewed efforts to restore legitimacy and rebuild trust in our information systems—provided we recognize the deeper systemic issues at play rather than focusing exclusively on technological solutions.