AI Scam Statistics 2026
By Axis Intelligence Research
Co-author: Marcus Chen | Last updated: November 5, 2026 | License: CC BY 4.0
Americans filed 22,364 complaints that mentioned artificial intelligence with the FBI in 2025 and reported $893,346,472 in losses, according to the FBI’s Internet Crime Complaint Center (IC3). That is 2.22% of all complaints but 4.28% of all reported losses, so a scam that involves AI costs its victim about twice as much as the average one.
Quick Answer
The FBI recorded $893.3 million in AI-related scam losses in 2025, its first year publishing an AI tag. Total US reported fraud losses reached $20.877 billion. Axis Intelligence Research finds that total is running at 0.97 of the path to Deloitte’s $40 billion-by-2027 generative AI fraud forecast, and victims aged 60 and over lost $112,153 per AI-related complaint.
Key Findings
- The FBI’s IC3 logged 22,364 AI-related complaints and $893,346,472 in AI-related losses in 2025, its first annual report with an AI descriptor.
- Axis Intelligence Research finds US reported fraud losses reached 0.97 of Deloitte’s generative AI fraud path in 2025: $20.877 billion reported against $21.43 billion on the curve to $40 billion by 2027.
- Axis Intelligence Research finds Americans aged 60 and over lost an average of $112,153 per AI-related FBI complaint in 2025, 3.99 times the $28,139 average for all other AI-related complaints.
- Axis Intelligence Research finds AI-related FBI complaints with a cryptocurrency link averaged $85,129 in losses in 2025, 7.66 times the $11,112 average for AI-related complaints without one.
- Of seven national fraud reports covering 2025 reviewed by Axis Intelligence Research, only the FBI’s IC3 report publishes a loss figure for AI-enabled scams (snapshot as of November 5, 2026).
How Much Money Is Lost to AI Scams?
The answer depends on who is counting, and the official count is a floor. The FBI’s figure is the only government number that isolates AI. Every other estimate is either a survey, a vendor’s telemetry, or a forecast.
The FBI’s AI line item
The FBI IC3 2025 Internet Crime Report tags a complaint as AI-related when “information reported contains a reference to artificial intelligence.” It counted 22,364 such complaints with $893,346,472 in losses, out of 1,008,597 complaints and $20.877 billion in total losses. The average AI-related complaint cost $39,946, which is 1.93 times the $20,699 average across all complaints.
The report also states that victims lost more than $5 million in 2025 to “distress” scams, the family-emergency calls where a cloned voice does the work. The full voice-clone breakdown, including the IC3 categories most exposed to synthetic speech, lives on our voice cloning scam statistics page.
What the AI tag misses
The IC3 tag only fires when the victim knows, or suspects, that AI was involved and says so. A cloned voice that sounded like a real nephew gets filed as a distress scam with no AI reference at all.
Two independent measurements show how wide that gap can be:
| Measure | Value | Basis | Source |
|---|---|---|---|
| Share of FBI complaints referencing AI, 2025 | 2.22% | Complaints filed with IC3 | FBI IC3 (Axis calculation) |
| Share of 2025 scams that involved AI or a deepfake | 12% | Survey of 5,173 US adults, Jan 8–Feb 18, 2026 | Gallup and Stop Scams Alliance |
| Scam victims who reported to the FTC or federal law enforcement | 13% | Same survey | Gallup and Stop Scams Alliance |
| Share of frauds reported to authorities in Canada | 5% to 10% | Agency estimate | Canadian Anti-Fraud Centre |
The Stop Scams-Gallup report found that 12% of 2025 scams reported to its survey involved AI or a deepfake, and that only 13% of victims went to the FTC or federal law enforcement. The two AI shares are different measures (scams in a household survey versus complaints in a police database), so Axis does not merge them. Side by side, the survey share is about 5.4 times the FBI’s tag rate, and both rely on the victim noticing the AI.
Reported losses versus estimated losses
| Figure | Value | What it measures | Source |
|---|---|---|---|
| FBI IC3 reported losses, 2025 | $20.877 billion | Complaints to the FBI | FBI IC3 |
| FTC reported fraud losses, 2025 | $15.9 billion | 3 million reports to Consumer Sentinel | FTC |
| Estimated scam losses, 2025 | $68 billion (95% CI: $33B–$114B) | Survey-based estimate, adults 18+ | Gallup and Stop Scams Alliance |
| FTC estimate adjusted for underreporting, 2024 | $31.3 billion to $195.9 billion | All ages, two methods | FTC |
| FTC reported fraud losses, 2024 | $12.5 billion | Reports to Consumer Sentinel | FTC |
The FTC told the Joint Economic Committee in March 2026 that consumers reported $15.9 billion in fraud losses in 2025 across 3 million reports. Its December 2025 report to Congress estimated that 2024 losses, adjusted for people who never report, ran between $31.3 billion and $195.9 billion for all ages. That is 2.5 to 15.67 times the $12.5 billion reported for 2024. The FTC calls the range highly uncertain, and Axis treats it that way: no multiplier from it is applied to the AI figures on this page.
Marcus Chen: Nobody should read $893 million as the size of the AI scam economy. Read it as the slice where the victim saw the machine. The interesting part is that even this self-reported slice costs twice the average complaint. The scams people recognize as AI are the expensive ones, which tells you where the tooling is being spent: on long investment cons, not on one-off text messages.
Is AI Fraud on Track to Hit $40 Billion by 2027?
So far, yes. US reported fraud losses are within 3% of the curve. The most-quoted AI fraud forecast on the internet is Deloitte’s “$40 billion by 2027,” and it is routinely misquoted as a global figure. It is a US forecast.
What Deloitte actually forecast
The Deloitte Center for Financial Services wrote in May 2024 that generative AI could push US fraud losses from $12.3 billion in 2023 to $40 billion by 2027, a 32% compound annual growth rate. Its model scored the fraud types tracked by the FBI’s IC3. In its aggressive scenario, generative AI email fraud alone reaches about $11.5 billion by 2027.
Two details matter for anyone citing it. First, the base is total fraud losses, not “AI fraud” in isolation: the forecast says generative AI will accelerate the whole category. Second, 32% a year from $12.3 billion reaches $37.34 billion in 2027; the rounded $40 billion headline implies about 34.3% a year.
The Generative Fraud Tracking Ratio (GFTR™)
The Generative Fraud Tracking Ratio (GFTR™) is an Axis Intelligence Research metric that checks the forecast against what the FBI actually records each year.
Formula: GFTR = IC3 reported losses for year t ÷ [$12.3 billion × 1.32^(t − 2023)]
A reading of 1.00 means US losses are exactly on Deloitte’s 32% path. Above 1.00, losses are running ahead of the forecast; below 1.00, behind it.
| Year | IC3 reported losses | Deloitte 32% path | GFTR™ | Source |
|---|---|---|---|---|
| 2023 | $12.5 billion | $12.30 billion | 1.02 | FBI IC3; Deloitte; Axis calculation |
| 2024 | $16.6 billion | $16.24 billion | 1.02 | FBI IC3; Deloitte; Axis calculation |
| 2025 | $20.877 billion | $21.43 billion | 0.97 | FBI IC3; Deloitte; Axis calculation |
| 2026 | Due in the 2026 IC3 report | $28.29 billion | Pending | Deloitte; Axis calculation |
| 2027 | Due in the 2027 IC3 report | $37.34 billion | Pending | Deloitte; Axis calculation |
US losses ran slightly ahead of the forecast for two years, then slipped just behind it. IC3 losses grew 25.8% in 2025 (the FBI rounds it to 26%), short of the 32% the curve assumes.
What 2026 and 2027 have to show
To reach $40 billion in 2027 from 2025’s $20.877 billion, US reported losses need to grow 38.4% a year. To stay on Deloitte’s 32% path ($37.34 billion), they need 33.7%. If 2025’s 26% growth simply repeats, IC3 losses land near $26.3 billion for 2026 and $33.1 billion for 2027. That is scenario arithmetic, not an Axis forecast.
The 2026 reading turns on one number: $28.29 billion. If the FBI’s 2026 report shows less, the GFTR falls further below 1.00 and the $40 billion headline needs a faster 2027 than any year IC3 has recorded in this series.
Marcus Chen: The forecast has held up better than most. Investment fraud alone added $2.1 billion in IC3 losses between 2024 and 2025. The question for 2027 is not whether AI makes fraud worse. It is whether reporting keeps pace with the losses, because a GFTR that drifts down could mean fewer victims or just fewer victims who file.
Which AI Scams Cost Victims the Most?
Investment fraud carries 70.75% of all AI-related losses, and business email compromise (BEC) produces the largest loss per AI case. BEC is a scam where criminals impersonate an executive or supplier to redirect a payment.
| Crime type (IC3) | AI-related complaints | AI-related losses | AI tag rate | Avg loss, AI-related | Avg loss, all complaints | Source |
|---|---|---|---|---|---|---|
| Investment | 4,356 | $632,041,188 | 5.97% | $145,097 | $118,500 | FBI IC3; Axis calculation |
| Business email compromise | 135 | $30,256,592 | 0.55% | $224,123 | $123,005 | FBI IC3; Axis calculation |
| Tech/customer support | 574 | $19,457,078 | 1.20% | $33,897 | $44,664 | FBI IC3; Axis calculation |
| Confidence/romance | 626 | $19,041,653 | 2.70% | $30,418 | $40,126 | FBI IC3; Axis calculation |
| Government impersonation | 260 | $7,061,628 | 0.80% | $27,160 | $24,610 | FBI IC3; Axis calculation |
| Extortion | 1,764 | $2,940,642 | 1.98% | $1,667 | $1,374 | FBI IC3; Axis calculation |
AI tag rate = AI-related complaints ÷ all complaints of that crime type.
Here’s what actually happened in that table. AI shows up most often where the con runs for weeks: 5.97% of investment complaints mention it, against 0.55% for BEC. Investment scams with an AI reference cost 1.22 times the average investment complaint, and BEC cases with one cost 1.82 times the average BEC case.
Romance and tech-support scams break the pattern. AI-related complaints in both categories average about 0.76 times the category norm. In those schemes AI appears to work as a volume tool (more chats, more fake support pages) rather than as the lever that pries loose a six-figure transfer.
Phishing tells a similar story: 803 AI-related phishing complaints carried $10,283,732 in losses. The lure-to-loss economics of phishing are broken out on our phishing statistics page.
Who Loses the Most to AI Scams?
Older Americans and anyone paying in crypto. Both groups lose several times more per AI-related complaint than everyone else.
| Segment, AI-related complaints, 2025 | Complaints | Losses | Avg loss per complaint | Source |
|---|---|---|---|---|
| Age 60 and over | 3,143 | $352,496,231 | $112,153 | FBI IC3; Axis calculation |
| Under 60 or age not stated | 19,221 | $540,850,241 | $28,139 | FBI IC3; Axis calculation |
| With a cryptocurrency link | 8,712 | $741,639,787 | $85,129 | FBI IC3; Axis calculation |
| Without a cryptocurrency link | 13,652 | $151,706,685 | $11,112 | FBI IC3; Axis calculation |
Americans aged 60 and over filed 14.05% of AI-related complaints but absorbed 39.46% of AI-related losses. Their average AI-related loss, $112,153, is 2.91 times the $38,500 average loss for all complaints from people 60 and over, per the FBI IC3.
Crypto is where AI scam money concentrates: complaints with a cryptocurrency link made up 38.96% of AI-related complaints but 83.02% of AI-related losses. Total crypto-linked losses reported to IC3 reached $11.37 billion in 2025, a picture covered in depth on our crypto crime statistics page.
Marcus Chen: Put those two rows together and you have the typical high-loss AI case in 2025: a retiree, an investment pitch, a convincing face or voice on the other end, and payment rails that can’t be recalled. Wire transfers can sometimes be frozen: the FBI’s Recovery Asset Team froze $679 million of $1.16 billion in attempted thefts it worked last year. Crypto sent to a scam wallet doesn’t go through that recall channel.
How Are AI Tools Changing Crypto Scams?
Scams that bought AI tools earned 4.5 times more per operation. Chainalysis traced on-chain payments from scam operations to AI vendors that sell face-swap software, deepfake services, and large language models.
| Chainalysis measure, 2025 | Scams linked to AI vendors | Scams without that link | Source |
|---|---|---|---|
| Average revenue per operation | $3.2 million | $719,000 | Chainalysis |
| Median daily revenue | $4,838 | $518 | Chainalysis |
Chainalysis estimates crypto scams took in at least $14 billion on-chain in 2025 and projects the final figure will exceed $17 billion as more wallets are identified. Impersonation scams grew more than 1,400% year over year, and the average scam payment rose from $782 in 2024 to $2,764 in 2025.
The 4.5x multiple is the strongest dataset-level evidence so far that AI raises the revenue of a scam operation, not just its volume. It is measured on blockchain flows, so it covers crypto-paid scams only.
Can Deepfakes Beat Identity Checks and Call Centers?
Deepfakes now account for one in five biometric fraud attempts, and contact-center data shows AI fraud growing about six times faster than ordinary fraud. The attack surface has moved from the inbox to the onboarding selfie and the phone line.
| Defender dataset | Figure | Period | Source |
|---|---|---|---|
| Share of biometric fraud attempts using deepfakes | 1 in 5 | Sep 2024–Sep 2025 | Entrust (company-reported) |
| Growth in deepfaked selfies | 58% | 2025 | Entrust |
| Growth in injection attacks | 40% year over year | 2025 | Entrust |
| Digital forgeries as a share of document fraud | 35% | 2025 | Entrust |
| Growth in “sophisticated” identity fraud | 180% vs 2024 | 2025 | Sumsub |
| Global identity fraud rate | 2.2% (2.6% in 2024) | 2025 | Sumsub |
| Growth in AI-driven fraud in voice channels | 1,210% (traditional fraud: about 195%) | 2025 | Pindrop (company-reported) |
| Targeted US consumers defrauded by AI scams | 27% (19% a year earlier) | May 2025 survey | Sift |
An injection attack feeds a pre-recorded or synthetic video stream straight into a verification app, bypassing the phone camera.
Entrust bases its figures on more than 1 billion identity verifications across 195 countries. Entrust sells identity verification, so it has an interest in the threat; it also runs one of the larger verification datasets in the field. Sumsub reports the opposite headline and the same story: the overall identity fraud rate fell to 2.2%, while sophisticated, multi-step fraud grew 180%. Fewer attempts, better ones.
Pindrop, which monitors contact-center calls, reported in March 2026 that AI-driven fraud grew 1,210% in 2025 against about 195% for traditional fraud. Pindrop does not disclose the sample behind the figure, so treat it as a directional signal from a vendor’s network. Sift’s May 2025 survey of 1,033 US adults found 27% of people targeted by an AI scam were defrauded, up from 19% a year earlier.
The case that set the price of a deepfake
The best-documented deepfake video-call loss is still Arup’s: an employee in the engineering firm’s Hong Kong office sent HK$200 million (about US$25.6 million) across 15 transactions after a video call in which the chief financial officer and other colleagues were all deepfake re-creations. Arup notified Hong Kong police in January 2024 and confirmed to CNN in May 2024 that “fake voices and images were used.” In November 2024, the US Treasury’s Financial Crimes Enforcement Network issued alert FIN-2024-Alert004 after banks filed a rising number of suspicious activity reports describing deepfake identity documents.
Marcus Chen: The Arup call worked because every control assumed a face on screen was a person. The defender data above says the same assumption now fails at onboarding, at a rate of one in five biometric fraud attempts. The unglamorous fix is a callback on a known number before money moves. It costs two minutes and it doesn’t care how good the deepfake is.
Which Countries Track AI Scam Losses?
Almost none. Axis Intelligence Research reviewed seven national fraud reports covering 2025. Every one names AI as a risk; only the FBI’s IC3 publishes a loss figure for it.
| Reporting body | Jurisdiction | Reported 2025 losses | AI-specific loss figure published? | Source |
|---|---|---|---|---|
| FBI IC3 | United States | $20.877 billion | Yes: $893.3 million | FBI IC3 |
| Federal Trade Commission | United States | $15.9 billion | No | FTC |
| UK Finance | United Kingdom | £1.28 billion | No (AI named, not quantified) | UK Finance |
| National Anti-Scam Centre | Australia | A$2.18 billion | No (AI named, not quantified) | NASC |
| Singapore Police Force | Singapore | S$913.1 million | No | SPF |
| Hong Kong Police Force | Hong Kong | about HK$8.1 billion | No | HKPF |
| Canadian Anti-Fraud Centre | Canada | over C$704 million | No | CAFC |
Snapshot as of November 5, 2026, based on each body’s 2025 annual report or annual release. Losses are in local currency and are not comparable across rows: each body defines fraud and scams differently.
UK Finance’s Annual Fraud Report 2026 shows authorised push payment (APP) fraud, where victims are tricked into sending money themselves, rising 19% to £576.4 million in 2025, with record investment-scam losses of £221.5 million. Its foreword describes criminals using AI and deepfakes but gives no AI figure. Australia’s National Anti-Scam Centre reported A$2.18 billion across 481,523 reports, a 7.8% rise in losses, and cites AI only as a reason scams are harder to recognize. The Singapore Police Force, Hong Kong Police Force, and Canadian Anti-Fraud Centre annual releases publish no AI-specific totals.
That single-ledger fact explains why one FBI number is quoted everywhere. It is also the strongest argument for treating any “global AI scam losses” figure with suspicion: no government outside the US measures the thing directly.
Marcus Chen: A regulator can’t manage what its fraud ledger doesn’t have a column for. Expect that to change first in markets with mandatory bank reimbursement, where the money at stake gives banks a reason to tag the cause.
How Are Platforms and Regulators Responding?
Ad platforms are removing scam inventory by the hundreds of millions, and the EU’s deepfake labeling duty now applies.
- Google blocked or removed more than 602 million ads under its scam-related policies in 2025 and suspended more than 4 million scam-linked advertiser accounts, out of 8.3 billion ads blocked or removed for all reasons, per its 2025 Ads Safety Report.
- Executives now rank cyber-enabled fraud ahead of ransomware: in the World Economic Forum’s Global Cybersecurity Outlook 2026, 73% of 804 surveyed leaders said they or someone they knew was directly affected by cyber-enabled fraud in 2025.
- The EU AI Act’s Article 50 requires providers to mark AI-generated audio, image, video, and text in a machine-readable way, and requires anyone deploying a deepfake to disclose it. It has applied since August 2, 2026, under Article 113 of Regulation (EU) 2024/1689. The AI omnibus, which the Council of the EU approved on June 29, 2026, gives providers until December 2, 2026 to put marking solutions in place, so 2027 is the first full year in which AI-generated content in the EU must carry a machine-readable label.
- Public trust keeps eroding: 61.6% of adults surveyed across markets in 2026 said they were concerned about what is real or fake on the internet, the highest reading since the series began at 54.2% in 2018, according to Reuters Institute survey data averaged in DataReportal’s Digital 2027 Global Overview Report.
US rules on AI voices in robocalls and the FTC’s impersonation rule are covered on our voice cloning scam statistics page; deepfake abuse laws and takedown deadlines are covered on our deepfake statistics page. For the breach side of AI, including the cost of AI-enabled intrusions, see our AI cyberattack statistics. If you need the practical version (how each AI scam works and what to do if you were hit), our AI scams guide walks through it.
Methodology
Collection. Every figure on this page was fetched from its producer between November 4 and November 5, 2026, and logged with its URL and retrieval date in the dataset below. Government sources (FBI IC3, FTC, UK Finance, NASC, Singapore Police Force, Hong Kong Police Force, CAFC, FinCEN) are used for reported losses. Company disclosures (Entrust, Sumsub, Pindrop, Sift, Google) are labeled as company-reported. Survey estimates (Gallup and Stop Scams Alliance, WEF) are labeled with their sample size and fieldwork dates.
Axis calculations. Averages per complaint divide IC3 losses by IC3 complaints for the same crime type, age band, or crypto segment. “Under 60 or age not stated” is all AI-related complaints minus the 60-and-over group, because IC3 publishes AI figures for only two age bands (60+ and 17 or younger). The AI tag rate divides AI-related complaints by total complaints of the same type.
GFTR™. The Generative Fraud Tracking Ratio divides the FBI IC3’s reported losses for a year by Deloitte’s 32% compound path from a $12.3 billion 2023 base. IC3 published its 2023 and 2024 totals as rounded figures ($12.5 billion and $16.6 billion), so those two readings carry rounding of about ±0.01. Deloitte’s base is about 1.6% below IC3’s 2023 total, which is why the 2023 reading starts at 1.02 rather than 1.00. The ratio tests the forecast against reported losses; it does not measure losses caused by AI.
What we chose not to do. Axis does not multiply the FBI’s AI figure by an underreporting factor. The FTC’s adjustment applies to all fraud, comes with a 2.5x-to-15.67x range, and was built for a different dataset. Survey figures and complaint figures are shown side by side and never averaged together.
National ledger review. For each of the seven bodies, Axis read the 2025 annual report or the official annual release and searched it for any loss, case, or report figure attributed to AI, deepfakes, or voice cloning. The snapshot is dated November 5, 2026, and will be rerun when 2026 reports are published.
About This Dataset
The dataset behind this page, ai-scam-statistics.csv, contains 261 rows: every figure in the article, the IC3 AI-related breakdown for all 26 crime types, the GFTR™ inputs and readings, and the national ledger review. Each row carries its source organization, document, URL, retrieval date, a primary-source flag, an Axis-calculated flag, and a method note for calculated rows.
It is updated when one of its inputs changes (a new IC3 or FTC annual report, a revised Chainalysis estimate), not on a fixed calendar. It is released under a CC BY 4.0 license: free to reuse with attribution to “Axis Intelligence Research, AI Scam Statistics 2026, 2026.”
How to Cite This Page
APA: Axis Intelligence Research. (2026, November 5). AI scam statistics 2026: The FBI counted $893 million, and the $40 billion 2027 forecast is still on schedule. Axis Intelligence. https://axis-intelligence.com/ai-scam-statistics/
MLA: Axis Intelligence Research. “AI Scam Statistics 2026: The FBI Counted $893 Million, and the $40 Billion 2027 Forecast Is Still on Schedule.” Axis Intelligence, 5 Nov. 2026, axis-intelligence.com/ai-scam-statistics/.
Chicago: Axis Intelligence Research. “AI Scam Statistics 2026: The FBI Counted $893 Million, and the $40 Billion 2027 Forecast Is Still on Schedule.” Axis Intelligence, November 5, 2026. https://axis-intelligence.com/ai-scam-statistics/.
AI Scam Questions Investigators, Banks and Families Ask
Is the “$40 billion in AI fraud by 2027” figure global?
No. Deloitte’s forecast covers the United States only and projects total fraud losses accelerated by generative AI, from $12.3 billion in 2023 to $40 billion in 2027. Axis Intelligence Research’s GFTR™ puts 2025 US reported losses at 0.97 of that path.
Why is the FBI’s AI scam number so much smaller than other estimates?
The FBI tags a complaint as AI-related only when the victim mentions AI, so the $893.3 million figure excludes cases where a clone went unnoticed. In a 2026 Gallup survey, 12% of 2025 scams involved AI or a deepfake, and only 13% of victims reported to the FTC or federal law enforcement.
Which AI scam causes the biggest losses per victim?
AI-related business email compromise averaged $224,123 per FBI complaint in 2025, the highest of any crime type. AI-related investment fraud averaged $145,097 per complaint and accounts for 70.75% of all AI-related losses, according to Axis calculations from FBI IC3 data.
Do AI scams hit older people harder than younger people?
Yes, by about four times per case. FBI IC3 data shows Americans 60 and over lost $112,153 on average per AI-related complaint in 2025, against $28,139 for all other AI-related complaints. They filed 14.05% of AI complaints but absorbed 39.46% of the losses.
Can a bank recover money lost to a deepfake or voice-clone scam?
Sometimes, if the payment was a wire and it is reported fast. The FBI’s Recovery Asset Team froze $679,013,183 in 2025, a 58% success rate on the thefts it worked. That process applies to bank transfers, and 83.02% of AI-related losses reported to the FBI in 2025 involved crypto.
Do deepfakes get past identity verification?
Some do. Entrust reports that deepfakes made up one in five biometric fraud attempts it saw between September 2024 and September 2025, and that injection attacks, which feed synthetic video directly into verification apps, grew 40% year over year.
Does any country besides the US publish AI scam loss figures?
Not in the 2025 reports Axis Intelligence Research reviewed. Of seven national fraud reports (FBI IC3, FTC, UK Finance, Australia’s NASC, Singapore Police Force, Hong Kong Police Force, and the Canadian Anti-Fraud Centre), only the FBI’s IC3 quantifies AI-enabled losses.
What will tell us whether AI fraud is accelerating in 2026?
The FBI’s 2026 Internet Crime Report. To stay on Deloitte’s path, US reported losses need to reach $28.29 billion for 2026, up from $20.877 billion in 2025. A lower figure would push the GFTR™ further below 1.00.