The FBI's Internet Crime Complaint Center marked its 25th year of operation by publishing a report that, for the first time, isolates the financial toll of artificial intelligence tools inside cyber-enabled fraud, attributing more than $893 million in losses to schemes referencing AI across 22,364 complaints filed in 2025.

The figures sit inside a broader accounting problem. Total losses reported to IC3 crossed $20.877 billion in 2025, a 26 percent jump from the prior year, spread across 1,008,597 complaints, according to the bureau. That volume translates to nearly 3,000 complaints arriving daily, a pace the center says has grown steadily since it opened in May 2000 with a caseload measured in the low thousands per month.

What distinguishes the 2025 edition from prior annual reports is the decision to track AI as a standalone descriptor rather than folding it into general fraud categories. Jose A. Perez, Operations Director for the FBI's Criminal and Cyber Branch, signed the report's opening section, in which the bureau describes artificial intelligence as a tool that "can be used for legitimate, helpful purposes or for criminal motives." The report frames the technology's dual nature bluntly: video and audio manipulation is not new, but the barrier to producing convincing fakes has collapsed.

2025_IC3Report
2025_IC3Report.pdf
3 MB
download-circle

How investigators traced AI into the fraud numbers

IC3 analysts applied the AI descriptor to a complaint only when the filer referenced artificial intelligence directly in their submission, meaning the true scale of AI involvement in fraud is almost certainly undercounted. The report itself acknowledges this gap when discussing investment fraud, noting that AI-nexus losses within that category reached $632,041,188, while total investment fraud losses exceeded $8.6 billion. That gap, the bureau states, demonstrates "that many victims do not realize the extent AI may be involved in scams."

Investment fraud carried the largest share of AI-referenced complaints by both count and dollar value. IC3 recorded 4,356 investment complaints citing AI tools, resulting in losses of $632,041,188. Business email compromise schemes involving AI generated $30,256,592 in losses. Confidence and romance scams with AI elements cost victims $19,041,653. Tech and customer support fraud citing AI reached $19,457,078. Personal data breach complaints referencing AI totaled $18,767,964.

The mechanics described in the report separate into a handful of recurring patterns. In business email compromise cases, chat generators can produce official-sounding emails that mimic a company's chief executive or other officials, complete with phishing links or wire instructions, while voice cloning can supplement the same request with a fabricated phone call. Confidence and romance schemes use AI chat generators to script fake profiles, making the resulting conversations more believable, and the report singles out so-called distress scams, or grandparent scams, in which voice cloning mimics a family member in an emergency. Those distress scams alone accounted for more than $5 million in claimed losses during 2025.

Employment fraud shows a different signature

Unlike the financially driven categories, AI-related employment fraud does not appear to chase large dollar figures. The report describes candidates using voice spoofing or deepfake video during online job interviews, where lip movement fails to sync cleanly with speech, or where physical cues such as coughing do not match what appears on screen. Victims reported losses of almost $13 million to AI-involved employment scams in 2025, a comparatively modest figure the bureau attributes to a different criminal objective: gaining access to private computer networks rather than extracting direct payments.

Investment clubs represent a related but distinct vector. According to the report, these operations employ AI-generated videos and voices of celebrities, executives, or other trusted figures to construct fraudulent, high-stakes opportunities, often featuring professional-looking endorsements distributed across social media or video calls. The report references a separate public service announcement, issued July 3, 2025, describing fraudsters targeting United States stock investors through investment clubs accessed on social platforms and messaging applications.

Cryptocurrency remains the dominant transaction layer

Underneath the AI figures sits a larger cryptocurrency problem that the report treats as the single costliest crime descriptor tracked in 2025. Complaints referencing cryptocurrency reached 181,565, a 21 percent increase from 2024, with associated losses of $11.366 billion, up 22 percent year over year. The average loss among cryptocurrency-linked complaints reached $62,604, and 18,589 complainants reported losing more than $100,000 each.

Cryptocurrency investment fraud specifically drew 61,559 complaints and $7.228 billion in losses, a 48 percent increase in complaint volume and a 25 percent increase in losses compared with 2024. The report attributes much of this activity to organized criminal enterprises based in Southeast Asia, describing operations that use victims of human trafficking as forced labor inside scam compounds. Contact typically begins through text messages, social media, advertisements, or dating applications before shifting to a private messaging platform, where victims are introduced to investment groups posing as industry insiders. Victims are shown fabricated profits and offered loans to encourage larger transfers, and when they attempt withdrawals, they face invented taxes and fees designed to extract further funds before the scheme operators disappear.

Cryptocurrency ATM and kiosk fraud, a narrower but fast-growing subset, drew 13,460 complaints and $389 million in losses, marking a 23 percent rise in complaint volume and a 58 percent jump in losses from 2024. Recovery scams, in which fraudulent firms claim to help victims retrieve funds lost in earlier schemes, added 10,516 complaints and $1.4 billion in reported losses, with the oldest age bracket, 60 and above, accounting for $540,505,960 of that total.

Older Americans absorb a growing share of the damage

Complainants aged 60 and older filed 201,266 complaints in 2025, a 37 percent increase from 2024, with associated losses of $7.748 billion, up 59 percent year over year. The average loss within this age group reached $38,500, and 12,444 complainants in the bracket reported losing more than $100,000. Investment fraud losses among this group alone reached $3,519,296,354, while tech and customer support fraud added $1,040,730,043 and business email compromise contributed $568,048,472.

Within the AI descriptor specifically, complainants 60 and older filed 3,143 AI-related complaints in 2025, a category the report notes was not tracked in 2024, making year-over-year comparison unavailable. Adjusted losses for this cohort under the AI descriptor reached $352,496,231, a substantial share of the $893 million national total attributed to AI-referenced fraud.

Sextortion complaints, which the report treats separately from the broader AI or elder fraud sections, totaled more than 75,000 in 2025, with IC3 referring more than 5,700 submissions involving minors to the National Center for Missing and Exploited Children. Among complainants 17 or younger, extortion was the single largest crime type by count at 5,151 complaints, and the report flags an ongoing FBI investigation into a violent online group it identifies as "764," which the bureau says coerces children into self-harm and other acts streamed live, in some cases resulting in the victim's death.

Government and law enforcement response

The report credits Operation Level Up, an initiative launched in January 2024 with support from FBI agents and the United States Secret Service, with notifying more than 8,000 cryptocurrency investment fraud victims and preserving more than $500 million in savings since its start. In 2025 specifically, the operation notified 3,780 victims, of whom 78 percent were unaware they were being scammed, for an estimated savings of $225,871,319. The report also states that 38 victims identified through the operation were referred to a victim specialist for suicide intervention, and describes specific interventions, including stopping one victim from cashing out $750,000 from a 401k retirement account and another from selling a house to send $500,000 to a scammer.

A newly formed Scam Center Strike Force, organized under the United States Attorney's Office for the District of Columbia, combines the Department of Justice's Criminal Division, the FBI, and the Secret Service to pursue the scam compounds operating in Cambodia, Laos, and Burma, according to the report. The Strike Force is also working with the State Department, the Treasury Department's Office of Foreign Assets Control, and the Department of Commerce to disable the internet infrastructure and social media access that supports these operations from within the United States.

The IC3 Recovery Asset Team, which coordinates with financial institutions to freeze fraudulent transfers through what the bureau calls the Financial Fraud Kill Chain, initiated 3,900 incidents in 2025 covering $1,163,919,846 in attempted theft, of which $679,013,183 was frozen, for a 58 percent success rate. Historically, the report notes, most of these interventions targeted business email compromise; in 2025, tech support and account takeover cases became more prominent, with some account takeover incidents involving 50 or more simultaneous transactions across multiple banks.

What the numbers leave unresolved

The report's own framing acknowledges a measurement limitation that matters for anyone assessing AI's true footprint in fraud: the AI descriptor is applied only when a complainant references the technology explicitly, and the investment fraud category shows the starkest example of the resulting gap, where AI-tagged losses represent roughly 7 percent of the category's total dollar losses despite the report's own acknowledgment that AI likely plays a role in a much larger share of cases. Whether that gap narrows in future reporting cycles will depend partly on whether victims themselves become better able to recognize AI-generated content when they encounter it, a challenge the bureau does not claim to have solved.

Ransomware, tracked separately from the AI figures, drew more than 3,600 complaints and losses exceeding $32 million in 2025, with 63 new variants identified through IC3 submissions at an average pace of 5.25 new variants per month. The top 10 reported variants, including Akira, Qilin, and Lockbit, accounted for 56.8 percent of all ransomware incidents and were concentrated most heavily in critical manufacturing, healthcare, and government facilities.

Timeline

  • May 2000: The FBI establishes the Internet Crime Complaint Center to receive reports of cyber-enabled fraud and cyber threats.
  • January 2024: Operation Level Up launches with support from FBI agents and the United States Secret Service to identify and notify cryptocurrency investment fraud victims.
  • March 6, 2025: IC3 publishes a public service announcement on violent online networks targeting vulnerable and underage populations.
  • July 3, 2025: IC3 publishes a public service announcement on fraudsters targeting United States stock investors through social media investment clubs.
  • August 11, 2025: The Department of Justice announces coordinated disruption actions against the BlackSuit, also known as Royal, ransomware operation.
  • November 2025: The FBI San Diego Elder Justice Task Force executes federal and state arrest warrants targeting an international elder scam network tied to more than 500 suspected victims.
  • December 2025: The FBI Baltimore Field Office and India's Central Bureau of Investigation dismantle a transnational cybercrime network based in Noida, India, arresting six individuals.
  • 2025 (full year): IC3 receives 1,008,597 complaints reporting $20.877 billion in total losses, including 22,364 complaints and $893,346,472 in losses tied to AI descriptors.

Summary

Who: The FBI's Internet Crime Complaint Center, under the Criminal and Cyber Branch led by Operations Director Jose A. Perez, published the findings, drawing on complaints filed by the public plus coordinated work with the United States Secret Service, the Department of Justice, the Central Bureau of Investigation in India, and international law enforcement partners.

What: The 2025 Internet Crime Report documents 1,008,597 complaints and $20.877 billion in total losses, with a new descriptor isolating 22,364 complaints and $893,346,472 in losses specifically referencing artificial intelligence tools, spanning investment fraud, business email compromise, romance scams, employment fraud, and tech support schemes.

When: The report covers calendar year 2025 and was published today, coinciding with IC3's 25th anniversary since its May 2000 founding.

Where: Complaints originated across all 50 states plus American territories, with California, Texas, and Florida recording the highest counts, and IC3 additionally received complaints from more than 200 countries outside the United States.

Why: The report matters because it represents the first year IC3 has tracked artificial intelligence as a discrete factor in cyber-enabled fraud, offering the advertising and marketing industry concrete figures on how generative tools, voice cloning, and synthetic media are being weaponized against consumers, a trend that intersects directly with platform trust and safety enforcement, ad verification standards, and the growing legal scrutiny of AI-generated commercial content documented across prior PPC Land reporting.