AI Lies Well: The Day Fake Voices and Faces Shake the Truth

Last time, we looked at the traps beyond the screen. Fake emails, a lover inside the screen, the lie of investment that crosses borders. Deception spread through the vast hunting ground of the internet, changing its form. Yet there still remained one premise. Text and images can be faked endlessly, but a living voice and face cannot be so easily faked—for a long time, that is what we believed.

That final premise is now being shaken. Artificial intelligence has become able to synthesize a voice exactly like the real thing and to move a face exactly like the real thing. The voice on the other end of the phone may not be your family’s. The superior shown in a video conference may not be your superior. The very boundary that separated lie from truth has begun to blur. In this final chapter of the series, we look at this new phase brought by AI, and, at the end of the history of deception, quietly consider what we ought to hold in our hands.

The Technology That Makes Fakes: The Arrival of the Deepfake

The word deepfake is said to have come into wide use around 2017. Combining deep learning, meaning deep machine learning, with fake, meaning a counterfeit, the word first became known as a technology for replacing a face within footage with that of another person.

The technology itself was not necessarily born of malice. In filmmaking, research, and the world of expression, it can open up great possibilities. But, as is the case with every powerful tool, it also carried a side that could be abused. From released video and audio, by having it learn the features of a specific person’s face and voice, one can fabricate footage that makes that person say what they never said and do what they never did. Telling it apart from the real thing is reported to grow more difficult year by year.

The Day a Voice Was Stolen: Two Reported Incidents

Events showing that this threat is no longer fantasy have actually occurred.

One is an incident reported in 2019. According to overseas reports, an executive at a certain energy-related company received a phone call using an AI-synthesized voice that impersonated the head of the parent company, and, believing the instructions, is said to have wired a large sum. The voice on the phone so closely resembled that of the familiar superior, it is reported, that there was no reason to doubt the caller. It is held up as an early, symbolic case in which the voice, that most intimate piece of evidence, conversely became a tool for deceiving people.

The other is an incident said to have occurred in Hong Kong in 2024. According to reports, an employee at a certain multinational company took part in a video conference. On the screen were familiar colleagues, including the head office’s chief financial officer. Yet not a single real human was there; all of it, it is said, was a fake appearance synthesized by deepfake. The employee believed the instructions and is reported to have wired a huge sum of funds. It is thought the faces and voices may have been learned from publicly released footage. In disguising several people at once and even making a real-time exchange possible, this incident delivered a great shock.

What these incidents teach is the fact that the progress of technology has begun to outrun human judgment. Our brains are made to instinctively believe a familiar voice and a familiar face. That instinct has, for tens of thousands of years, done the important work of connecting person to person. The deceiver now targets that very instinct. We are entering an age in which we are asked to doubt the very thing that could not be doubted.

At the End of the History of Deception: Having an Eye That Sees Through

In this series, we have made a long journey. Beginning with the ancient deceptions left in the records, through the alchemy that claimed to make gold, the bubble that whipped crowds into a frenzy, the Ponzi scheme promised to collapse, the confidence trick that steals the heart, the counterfeit exactly like the real thing, the investment fraud said to be the largest in history, the special fraud carried out over the telephone, the internet fraud beyond the screen, and on to the age in which AI tells lies.

For all that it changed form, for all that it evolved this far, the essence of deception has remained astonishingly unchanged. The place the deceiver aims for is always the same. Human greed, human anxiety, human loneliness, and the heart that tries to believe in people. Technology is no more than a tool. The alchemist’s crucible, the telephone, AI—all were tools for slipping into the gaps in the human heart. And so the history of deception was, at the same time as being a history of technology, also a history of the weakness and the strength of the human heart.

The history of deception has no clean ending. As long as there are deceivers, the deceived, too, will likely never disappear. Each time a new technology appears, someone who abuses it appears, and we brace ourselves once again. This is an endless game of cat and mouse. We cannot lightly say that it is all right now.

Yet there is one thing we can do. It is to keep trying to have an eye that sees through. To know the history of deception is to know where in the human heart it is aimed. The person who knows the mechanism—why people come to believe, why they grow flustered—can, before the same trap, pause just a little. That single moment of pausing protects you, and protects someone dear. To doubt is not to disbelieve in people. Rather, it is to calmly try to discern what is worthy of belief.

To all the readers who have read this long-running journey this far, I give my heartfelt thanks. The story of the deceivers and the deceived will likely continue, changing form, from here on. But your eyes, which have watched that story, should no longer be the same as before. Please, keep quietly holding that eye that sees through, from here on as well. For it is the modest and certain wisdom that the long history of fraud has left to us.

【A World History of Fraud: Deceivers and the Deceived — End】