The fraud landscape has shifted significantly over the past two years, and the statistics highlight this change clearly. According to the FBI Internet Crime Complaint Center, reported losses reached a record $20.9 billion in the latest annual report, marking a 26% increase in just one year, from over one million complaints.
Globally, identity fraud losses exceeded $50 billion in 2025 alone, and early signs point to an even worse scenario for 2026. The main cause isn’t just an increase in scammers but also the availability of better tools. Artificial intelligence has quietly become the dominant force transforming financial fraud over the past generation, and most people remain unaware of how advanced it already is.
Why Scams Are Multiplying So Fast
Financial fraud is not new, but the economics of running a scam have completely changed. Generative AI has collapsed the cost and skill needed to produce convincing fake identities, voices, videos and documents at scale.
A few numbers explain why:
- Deepfake-related fraud losses have already reached roughly $2.19 billion globally, with the US alone accounting for over $712 million.
- Document deepfakes, fake IDs, forged pay stubs, altered bank statements, are projected to rise nearly 3,900% year over year as generation tools become more accessible.
- Nearly half of businesses worldwide now report being targeted by audio or video deepfake fraud attempts.
- Deloitte projects that generative AI-enabled fraud losses in the US could hit $40 billion by 2027, up from around $12 billion in 2023.
Fraud researchers see this less as an increase in scams and more as a rise in scam sophistication. While the total number of complaints has remained fairly constant in some areas, the success rate of scams has increased significantly, from about 19% of targeted individuals being defrauded in 2024 to 27% in 2025. This is because scams themselves have become much more convincing.
The Dangerous Role AI Actually Plays
This is the part worth understanding clearly, because it is not abstract. AI is being used in a few very specific, very effective ways:
Voice cloning. Scammers only need about three seconds of audio, often pulled from a public TikTok, Instagram story or old voicemail, to build a convincing clone of someone voice. That clone is then used to call a family member, claiming to be in an accident, to have been arrested, or to be in danger, demanding urgent payment.
Deepfake video calls. This is no longer theoretical. In one widely reported case, a finance employee at an engineering firm transferred $25 million after joining what appeared to be a video call with the company CFO and colleagues. Every person on that call was an AI-generated deepfake.
AI-written phishing and investment scams. Generic, typo-filled phishing emails are being replaced by AI-generated messages tailored to a specific person’s job, writing style, and even their real coworkers’ names, pulled from public data and social media.
Celebrity and executive impersonation. More than half of all deepfake fraud losses worldwide now stem from fake celebrity or government official endorsements used to promote fraudulent investment schemes.
Humans are generally not equipped to recognize this. Studies on high-quality deepfake videos reveal that people identify them as fake only around 24.5% of the time. Additionally, despite common assumptions, deepfake-detection apps are not entirely dependable. Even tools that achieve 94 to 96% accuracy in controlled lab settings often fall below 50% accuracy when tested against real-world tools they haven’t been trained on.
Who Is Actually Getting Hit
The pattern is not quite what most people expect. Younger, phone-native adults are targeted and defrauded more often; in one FTC dataset, 44% of people in their 20s targeted in 2024 lost money, compared with 24% of people in their 70s.
But older adults lose far more per incident. Americans over 60 reported around $7.7 billion in cybercrime losses in 2025 alone, a 59% jump from the year before, with a median loss of over $1,600 per victim, sometimes climbing into six figures for the worst cases.
The five sectors seeing the heaviest fraud rates right now are dating platforms, online media, financial services, crypto, and professional services, largely because these are industries where verifying someone identity is the entire transaction.
What Actually Works to Protect Yourself
Security experts are consistent on one point: you cannot out-listen or out-spot a good deepfake. The defense has to be procedural, not perceptual.
- Set a family code word. Agree on a word or phrase with close family members that a scammer could never guess. If someone calls claiming to be a relative in distress, ask for it before doing anything else.
- Never call back on the number that called you. If you get an urgent call, hang up and dial the person directly using a number you already have saved.
- Treat urgency and secrecy as red flags. Legitimate banks, agencies and family emergencies rarely require instant payment via gift cards, crypto, or wire transfer, and rarely ask you to keep the situation secret.
- Verify through a separate channel. For any unusual payment request, especially over video call, confirm it through a different, pre-established communication method before acting.
- Lock down your public audio and video. Limit how much voice and video content of yourself and your family is publicly accessible, since that is the raw material scammers use to build clones.
- Report it immediately if it happens. Contact your bank to attempt a freeze or reversal, then report the incident to the FTC at reportfraud.ftc.gov and the FBI IC3 at ic3.gov.
The Bottom Line
AI has not only increased the frequency of scams but has also made them more difficult to identify through traditional instinctive methods. Features such as voice tone, video presence, or personalized email details can no longer be trusted solely as proof of identity. The encouraging news is that the strongest defenses are simple and require no technical expertise.
Using a code word, confirming a call back from a known number, and maintaining a healthy suspicion of urgent requests are cost-free strategies that, as the FBI states, remain the most effective defenses against AI-driven scams.
