Prof. Dr. Dennis-Kenji Kipker is a cybersecurity expert and works as Scientific Director of the cyberintelligence.institute.getty​For as long as money has existed, fraud has followed it like a shadow, and the history of finance can be read as a long contest between those who deceive and those who try to detect them. Artificial intelligence now steps into this contest on both sides at once, because the same technology that promises to catch criminals before they strike also hands those same criminals tools of a power they have never enjoyed before. From my own experience in advisory work, I can say that few developments have reshaped the conversation in finance departments as quickly as this one, and understanding how the dual role unfolds matters for anyone who handles money, whether they sit behind the counter of a bank, run a small business or simply keep a modest savings account.When The Con Artist Learns To CodeThe most unsettling change is how convincingly a fraud can now be staged, since a scammer who once needed a gift for mimicry and a great deal of luck can today rely on software that clones a voice from only a few seconds of recorded speech. The first widely reported case of this kind reached the public in 2019, when criminals used an artificial voice to imitate the chief of a German parent company and persuaded the head of a British subsidiary to wire roughly $240,000 to a fraudulent account. When a finance officer receives an urgent telephone call from someone who sounds exactly like the chief executive, the temptation to act quickly can override every instinct of caution, so that the money may be gone before anyone thinks to question what was heard.What once required a single cloned voice has since grown into something far more elaborate, because the same generative systems now compose flawless messages in any language and even fabricate video calls in which several familiar faces appear to speak at once. In one striking case in early 2024, an employee at an engineering firm joined what looked like an ordinary conference call and authorized transfers of around $25 million, only to learn afterward that every colleague on the screen, including the CFO, had been an artificial recreation. In my personal assessment, this case marks the moment at which the old reassurance of having spoken to someone in person subtly lost much of its authority.Beyond the art of impersonation, artificial intelligence helps fraud operate at a scale that human criminals could never reach on their own, because a single operator can now direct software that shapes thousands of personalized messages, each one adjusted to the habits and the fears of the person who receives it. Identities that have been stitched together from a blend of real and invented details slip through checks that were designed to spot only the most obvious lies, and this form of deception, which is usually called synthetic identity fraud, has surged into one of the most serious threats in finance as generative tools accelerate both its scale and its realism. Because these systems learn from every failure, they grow steadier with each attempt, so that the fraud which fails today teaches the fraud that may succeed tomorrow.The Guardian Built From The Same ClothThe institutions that protect our money are nevertheless far from defenseless, since the very capacity for recognizing patterns that empowers the attacker also sharpens the eyes of the watchman. Traditional controls leaned on fixed rules that raised an alarm whenever a transaction crossed a set threshold, an approach that produced a flood of false alarms while it missed anything clever enough to stay just beneath the line. Artificial intelligence replaces this blunt instrument with a system that gradually learns what ordinary behavior looks like for each individual customer, so that a purchase made in an unfamiliar city at a strange hour can be weighed against a lifetime of habits rather than against a single rigid limit.Because these models digest far more information than any analyst could ever review by hand, they notice the faint signals that betray a deception even when each signal looks innocent on its own. Several large banks now report that such behavioral systems have sharply reduced the number of false alarms while catching more genuine fraud. From my experience in consulting practice, I can say that the institutions which gain the most are rarely those with the most expensive software, but rather those that feed their systems clean and well-connected data, because a model can only ever be as perceptive as the information it is given. When an account that has slumbered quietly for years suddenly springs into frantic activity, or when someone hesitates over a payment screen in a way that differs from their familiar rhythm, the system registers the oddity and can intervene within a fraction of a second, often before the victim realizes that anything has gone wrong at all.An Arms Race Without A Finish LineThe honest conclusion is that artificial intelligence grants no lasting victory to either side, because every advance in detection teaches the attacker something new, while every fresh deception trains the defender in turn. When fraudsters adopt voice cloning, banks respond with checks that test for the subtle traces cloning leaves behind, and when defenders turn to the study of behavior, criminals begin to rehearse how ordinary conduct can be imitated. The contest therefore accelerates rather than settles, and in my own assessment, the institutions that avoid falling victim will be those that treat the prevention of fraud as a living discipline rather than a puzzle to be solved once and then forgotten. What lies before us is not a world cleansed of fraud, but a world in which the speed of both deception and defense keeps rising, so that the responsibility shared between institutions and individuals grows more important. The oldest weakness that fraud exploits has always been human rather than technical.Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?