What total losses will the FBI's 2026 Internet Crime Report state were reported to IC3 for calendar year 2026?

TL;DR

The forecast is Median (50th percentile): $26.10 billion calendar year 2026. The uncertainty bounds are 10th percentile: $21.20 billion calendar year 2026; and 90th percentile: $32.05 billion calendar year 2026. The central expectation is another substantial increase in reported losses, with reporting coverage and large individual incidents driving the largest risks.

Context

The target is the FBI’s overall IC3 reported monetary-loss total for January 1 through December 31, 2026, expressed in nominal US dollars. It covers the complaints included in the report’s headline statistics, including international complaints where included. It is not a measure of all internet crime, an independently verified tally of financial harm, or a total net of recoveries. The official report’s designated aggregate controls, even if its coverage or methodology changes.

As of the November 5, 2026 forecast date, the annual-report index listed 2025 as the latest report, and no usable official all-category 2026 interim total had been identified (IC3 annual reports). Most of the reporting year had elapsed, but that does not mean most of the target was observable. Under the resolution rules, a qualifying report must be published by August 31, 2027; otherwise, the question is annulled.

Evidence

The historical backbone is a sustained rise in reported losses. The complete annual history available in the research runs from 2017 through 2025. The earlier observations come from the historical chart in the 2021 IC3 report, and the later sequence is documented in the 2025 IC3 report. These are published historical figures, with older values rounded as shown, rather than a reconstructed unrevised database.

Calendar year	Reported losses, USD billions
2017	1.4
2018	2.7
2019	3.5
2020	4.2
2021	6.9
2022	10.3
2023	12.5
2024	16.6
2025	20.877

The series supports continued growth more strongly than a plateau. But it does not show a fixed growth rate. The earlier years include much faster proportional increases, while the latest completed year grew approximately 26%. The forecast therefore rests on persistence in the recent upward trend, not on an assumption that the fastest historical growth continues (2025 IC3 report).

The latest increase also had more than one driver. Complaints rose from 859,532 in 2024 to 1,008,597 in 2025. Together with the loss totals, those figures imply roughly 17% complaint growth and 7% growth in losses per complaint. I read this as a broader source of momentum than complaint volume alone, though neither driver is guaranteed to persist (FBI release).

Category composition helps explain where changes can matter. In 2025, investment fraud accounted for $8.6486 billion, business-email compromise for $3.0466 billion, technology/customer support scams for $2.1347 billion, and government impersonation for $0.797943193 billion. Cryptocurrency and AI are overlapping descriptors, not separate amounts to add to these categories. A sharp increase in a small category need not move the aggregate as much as a modest change in investment fraud (2025 IC3 report).

The strongest specific partial-period US signal concerns impersonation. A September 17, 2026 advisory reports nearly 61,000 government/law-enforcement impersonation complaints and more than $1.6 billion in losses during January 2025–July 2026. Compared with the completed-year category total, this points toward continued pressure in 2026. It is not an official 2026 subtotal: the window crosses years, the reported amount is a rounded lower bound, and category alignment is unresolved. It supports the direction of the forecast without establishing aggregate acceleration (IC3 advisory).

Foreign comparators provide a useful check against a one-way growth narrative. Canada reported approximately C$351 million in fraud losses in the first half of 2026. A parliamentary record gives C$342 million through June 30, 2025, suggesting growth of about 3%, subject to snapshot and revision differences. These are Canadian fraud-reporting figures, not substitutes for IC3 observations (Canadian release; parliamentary evidence).

Singapore’s full first-half scam-loss history in the supplied research is S$342.1 million in 2022, S$309.4 million in 2023, S$522.4 million in 2024, S$500.2 million in 2025, and S$410.6 million in 2026. Its latest first-half total fell 17.9%; investment losses fell 5%, and crypto-payment losses fell 43%. Different reporting systems, payment definitions, and enforcement regimes limit direct comparison. Still, these observations show that rising losses are not universal (Singapore mid-year report).

Enforcement is another reason to retain a slowdown risk. The DOJ’s October release reports approximately $1.006 billion cumulatively restrained and 700 fraudulent investment websites seized. It also describes scam operations relocating across countries. The confirmed fact is substantial disruption; the inference is that prevention can restrain growth, while adaptation limits lasting effects. Restrained funds are not measured annual losses prevented and are not deductions from the target (DOJ release).

The Binance incident creates a different kind of risk: statistical coverage rather than ordinary fraud growth. The FBI attributed the October 29, 2026 theft of more than $6 billion to North Korea. Binance’s November 3 preliminary update estimated $6.34 billion affected and $2.85 billion frozen or intercepted, and said it would absorb the financial impact rather than users. Neither statement establishes how IC3 will treat the incident in its annual aggregate (FBI statement; Binance update).

A prior analogue cautions against automatic inclusion. The FBI publicized the approximately $1.5 billion Bybit theft in February 2025, but the supplied annual-report evidence does not explicitly reconcile that theft into its headline total. Public confirmation of an attack is not proof of inclusion in a complaint-based statistic (IC3 Bybit advisory).

AI and crypto research adds weak signals, not an aggregate growth measure. TRM reported rapid growth in narrow AI-scam indicators, but missing absolute amounts, sample sizes, and consistent windows prevent translating those rates into IC3-wide growth. Likewise, industry reports titled “2026” often describe 2025 activity. Blockchain attribution, community reports, and IC3 complaints measure different things; their totals cannot be treated as interchangeable (TRM AI report; Chainalysis scams report; TRM crypto crime report).

What's non-obvious

The biggest headline is not necessarily the biggest addition to the resolving statistic. A large exchange theft can dominate coverage without entering IC3’s headline in full. Conversely, more victims reporting ordinary fraud can lift the aggregate even if underlying criminal activity grows more slowly. FinCEN’s guidance encouraging referrals to IC3 is a potential reporting-uplift signal, but its effect is unquantified (FinCEN alert).

The same distinction applies to successful enforcement. Frozen assets, recoveries, estimated avoided losses, and annual reported losses answer different questions. Subtracting enforcement totals would create false precision. The useful reading is directional: disruption supports a slowdown risk, continued impersonation losses support growth, and foreign declines argue against assuming acceleration across every category and market.

Uncertainties

The largest gap is the unpublished all-category 2026 aggregate. A comparable official interim total, with clear coverage and revision notes, would do more to narrow uncertainty than additional isolated scam alerts. The absence of that total is a visibility problem, not evidence that losses have stalled.

The second gap is treatment of exceptional complaints, especially Binance. A clear statement of whether the incident enters the headline total, and on what basis, would resolve an important upper-tail uncertainty. Reimbursement and frozen assets alone do not answer that question.

Finally, the statistic itself can move through reporting participation, classification, and investigative adjustments. IC3 describes its statistics as point-in-time assessments and says losses and classifications can change, with duplicates removed as far as possible (2025 IC3 report). These uncertainties preserve meaningful scope for both a slowdown and an upside surprise. They do not change the saved forecast.