Deepfakes aren’t a fringe internet phenomenon. They are increasingly being embedded in familiar forms of fraud and financial manipulation, reshaping how scams look, sound, and feel.
At their core, deepfakes are synthetic media created using deep learning techniques to generate audio, video, images, or text that convincingly imitates real people or events. The technology combines advanced machine learning models with fabricated or manipulated information, producing content that appears authentic even when it’s entirely false.
The term “deepfake” itself emerged in 2017 from an online community experimenting with early face‑swapping tools, but the underlying techniques have a longer lineage. Computer‑generated imagery entered research labs in the 1960s, became visible to mainstream audiences through films like Westworld and Tron, and later evolved through digital morphing techniques made popular in music videos and cinema during the 1990s. The creation of what’s known as generative adversarial networks or GANs in the mid‑2010s marked a turning point, enabling computers to generate increasingly realistic human faces, voices, and movements.
“What has changed most dramatically in recent years is not just the quality of deepfakes, but their accessibility,” says Rich Graham, Moody’s industry practice lead. “High‑resolution video and audio of public figures, executives, and everyday individuals is widely available online, providing abundant training material for synthetic media tools.”
At the same time, the technical barrier to producing convincing fakes has been lowered. Tools that once needed specialists to develop them are now easier to access, meaning deepfake techniques can be leveraged by fraudsters at increasing speed and scale.
Moving way beyond the bounds of movie entertainment, deepfakes are now appearing as levers in established scam models. Impersonation scams, romance scams, investment fraud, charity fraud, and sextortion schemes pre‑date generative AI. But now deepfakes are being used to add a layer of perceived authenticity that can make these familiar schemes even harder for targets to recognize. Ironically, technology can be part of what makes the scams more emotionally compelling to human audiences.
The adoption of deepfakes in criminal enterprises is rising against a broader backdrop of increasing fraud losses. In the United States alone, the Federal Trade Commission reported that consumers disclosed losses of $12.5 billion to fraud in 2024, a 25% increase on the previous year. The rise in reported losses does not appear to stem from more fraud reports overall, but from scams becoming more effective. According to the FTC, a significantly higher proportion of reported cases in 2024 resulted in financial loss, particularly within imposter style scams that exploit assumed identity and credibility. Imposter scams were among the highest‑loss categories, highlighting how deception often relies on assumed identity and trust.
The increasing use of deepfakes in criminal enterprises is also reflected in the FBI's 2025 IC3 Report, which tracked more than $893 million in losses from complaints mentioning Artificial Intelligence (AI). These schemes included investment scams using AI-generated videos of public figures and distress scams using voice cloning to impersonate family members. This trend appears to be part of a larger surge in cyber-enabled fraud, with total reported losses reaching $20.9 billion in 2025, underscoring that deception remains a core tactic.
One of the most widely cited deepfake‑enabled fraud cases took place in Hong Kong in early 2024.
A finance employee at a multinational company was instructed to transfer funds after joining what appeared to be a routine internal video call. The participants on the call looked and sounded like the organization’s chief financial officer and other colleagues.
In reality, every participant except the victim was a synthetic reconstruction created using publicly available footage. Over the course of multiple transactions, approximately US$25.6 million was transferred before the fraud was uncovered.
This case illustrates how deepfakes can provide social “confirmation” that overrides initial skepticism, particularly when combined with urgency and perceived authority.
Another area where deepfakes can intersect with serious harm is sextortion and image‑based blackmail. The National Center for Missing & Exploited Children reported 26,718 cases of financial sextortion in 2023, more than double the number recorded in 2022. While not all these cases involve synthetic media, deepfakes can be used to strengthen false personas or to generate fabricated compromising material that increases coercive pressure.
In parallel, Meta disclosed it removed around 63,000 Instagram accounts in Nigeria linked to financial sextortion activity, along with thousands of associated Facebook pages and groups used to distribute scam scripts and techniques.
“Our market observations highlight the growing industrialization of scam operations. Investigations in Southeast Asia, for example, have identified reports of large‑scale scam centers that combine tactics like romance fraud, investment scams, and deepfake impersonation,” adds Graham.
In January 2026, South Korean authorities repatriated 73 nationals from Cambodia in connection with online scam operations, including allegations of deepfake‑enabled romance and investment fraud affecting hundreds of victims and tens of millions of dollars. These cases point to a broader ecosystem in which synthetic media is one tool among many, deployed within organized and often transnational criminal activity.
Underlying many of these scams is a human vulnerability that exploitation of technology alone cannot explain. The US Surgeon General has described loneliness and social isolation as a significant public health challenge, noting social connection plays a critical role in individual and societal wellbeing. Romance scams and impersonation fraud often prey on this, using attention, empathy, and perceived familiarity to build trust before introducing deception. Deepfakes can intensify this dynamic by making interactions feel more personal and more real.
Taken together, these developments help explain why deepfakes are increasingly being viewed as a mainstream risk rather than a technological curiosity. They are believable, repeatable, and easily integrated into existing fraud patterns. While detection tools and platform controls continue to evolve, deepfakes illustrate a broader shift in the risk landscape: seeing and hearing are no longer sufficient indicators of authenticity.
In that sense, deepfakes are best understood not as a single threat, but as a new layer within deception that reshapes how trust is constructed, tested, and sometimes exploited in the digital world.
Deepfakes highlight why context, verification, and correlation of risk signals matter in modern risk and compliance workflows. As synthetic media blurs the line between genuine and fabricated identities, organizations are increasingly looking beyond single data points to understand who they are dealing with, how entities and individuals are connected, and whether behavior aligns with known risk patterns.
Moody’s compliance and third-party risk management solutions bring together structured data, entity and people linkages, adverse media, sanctions, and other contextual signals to help organizations assess identity‑related risk within a broader operating environment.
By layering multiple sources of data and continuously enriching them with news and event context, Moody’s supports more informed decision‑making across onboarding, monitoring, and investigations, particularly as deception techniques, like deepfakes, become more sophisticated and harder to spot in isolation.
Please get in touch with the team at Moody’s to discuss your approach to compliance and risk management. We would love to hear from you.
*Disclaimer: This content is for informational purposes only and does not constitute legal, financial, compliance or other professional advice. Please consult with a qualified professional for specific legal, financial, compliance, or other professional advice. For more terms and conditions pertaining to Moody’s products and services, refer to the https://www.moodys.com/web/en/us/legal/global-disclaimer.html on Moody’s website.