American Thinker

by Delise Tattum

    New AI tools are already capable of generating highly convincing audio and video. Voices are being cloned with remarkable accuracy. Images are being fabricated that are almost indistinguishable from authentic photographs. And video technology has advancing to the point where a person can appear to say things that were never actually spoken.

    Artificial intelligence promises extraordinary advances in science, medicine, and technology. But in the realm of public discourse, it may push misinformation into an entirely new phase. Until recently, most distortions have relied on editing real events -- cutting clips, removing context, and rearranging fragments of genuine footage. AI makes something more radical possible: the creation of convincing events that never occurred at all.

    Public debate once rested on the assumption that public speech would be presented honestly and interpreted in good faith. Increasingly however, as many have realized, this is no longer the case.

    Over the past decade, the phrase “fake news” entered the political vocabulary to describe the manipulation of information in modern media. These distortions did not usually arise from elaborate fabrications, but from simple manipulations of context and presentation. They rarely involved completely invented stories. More often than not, they relied on the careful reshaping of reality: statements taken out of context, video clips selectively edited, narratives framed around fragments rather than the full record.

    That process has become all too familiar. A short clip circulates online. A headline cements the accusation. Commentators repeat the interpretation until it becomes the dominant narrative. By the time the fuller context emerges, the story has already taken hold.

    Several widely known episodes illustrate how this process works.

    After the 2017 Charlottesville protests, a fragment of Donald Trump’s remarks -- “very fine people on both sides,” referring solely to the Confederate statues debate -- circulated widely as evidence that he had praised white supremacists. Yet in the same speech he explicitly stated that neo-Nazis and white nationalists should be “condemned totally.” That portion of the speech was conveniently excluded in the viral clip.

    During the COVID-19 pandemic, the suggestion that the virus may have originated in a laboratory leak was widely dismissed as misinformation. Social media platforms restricted discussion of the possibility, sometimes removing posts or labeling them misleading because they conflicted with prevailing public-health guidance. Over time, however, the hypothesis came to be considered a plausible explanation by scientists and intelligence agencies.

    The Hunter Biden laptop story followed a similar trajectory. Reporting that appeared shortly before the 2020 U.S. presidential election widely described the story as Russian disinformation, and social media platforms limited the story’s distribution. Later reporting confirmed that the laptop and much of its content were authentic.

    Even international conflicts can be framed through selective representation. In debates surrounding military strikes on Iran, much Western commentary has focused primarily on geopolitical escalation. Yet the reactions of many Iranians -- particularly those in the diaspora who oppose the clerical regime established in 1979 -- often receive far less attention. When some voices are amplified while others are muted, audiences are left with an incomplete picture of reality.

    In each of these cases, the mechanism is similar. Fragments are arranged in ways that produce a misleading narrative. Selective editing functions as a form of narrative construction.

    These distortions relied on something real. A speech had to exist. A video had to be recorded. A quote had to be spoken, even if it was later taken out of context.

    Artificial intelligence has changed that.

    In the past, misleading clips could usually be challenged by locating the full recording of the event. But AI tools now allow users to create highly convincing altered versions of recordings, making it far harder for audiences to know which version is authentic.

    Ironically, the spread of deepfakes may even make genuine evidence easier to dismiss, as any damaging recording can simply be described as artificial.

    The deeper danger is not merely that false videos will circulate online. It is that genuine evidence may no longer be trusted. When audiences know that convincing fabrications are possible, even authentic recordings can be dismissed as artificial. The result is a world in which both truth and falsehood become indistinguishable.

    The paradox of the age of artificial intelligence is that the same technology capable of fabricating convincing versions of reality can also help verify authentic records. AI systems are increasingly being developed to detect manipulated images, cloned voices, and altered video by analyzing digital fingerprints, inconsistencies in pixels, lighting, and voice patterns. At the same time, emerging tools such as digital watermarking and systems that track the origin and history of recordings can help confirm when and where original material was created and whether it has been altered. But these tools will only matter if authentic records are preserved and accessible. If reliable source material disappears while AI-generated versions circulate freely, public understanding will increasingly be shaped by whichever narrative spreads most effectively.

    In this environment, the preservation of source material becomes critically important.

    Full speeches, unedited interviews, original documents, and primary recordings will increasingly serve as the anchors of public verification.

    In a world where AI-generated media can blur the line between authentic and fabricated events, the challenge facing democratic societies is not to eliminate disagreement or political conflict. Such conflict is an inevitable feature of free political life. The real task is to establish reliable methods of verification -- transparent systems that allow anyone to trace claims back to authenticated source material.

    In this new phase of fake news, the challenge is no longer determining whether a statement has been taken out of context.

    It is determining whether it was ever spoken at all.