AI in Finance Statistics 2026
By Axis Intelligence Research
Co-author: Sarah Davis | Last updated: July 1, 2026 | License: CC BY 4.
Quick Answer
Ninety percent of financial institutions now use AI to detect fraud, according to a 562-respondent global survey from Feedzai — yet only 56% of finance leaders use AI inside their own department, the lowest adoption rate of any business function, per a 2026 CFO Connect survey built on General Atlantic polling data. Axis Intelligence Research calls that 34-point split the Back-Office AI Gap.
AI use in financial services is sharply uneven by function and by firm size in 2026. Fraud-detection teams have all but finished adopting it: 90% of financial institutions use AI to fight fraud, and two-thirds did so within the past two years. Finance departments — the CFO’s own office — trail badly: just 56% report using AI at all. That figure roughly doubles to 76%+ once the sample narrows to large firms with $250 million or more in annual revenue, per KPMG’s March 2026 survey of 1,013 finance leaders. Meanwhile, McKinsey’s standing estimate puts generative AI’s addressable value in banking at $200 billion to $340 billion a year — a figure regularly confused, in wide circulation, with an unrelated $3.8 trillion number that actually describes total 2025 revenue at the world’s 1,000 largest banks.
Key Findings
- Axis Intelligence Research calculates a 34-percentage-point Back-Office AI Gap Index (BOAG™) between financial institutions’ near-universal adoption of AI for fraud detection (90%, Feedzai 2025) and finance departments’ own AI adoption (56%, CFO Connect 2026) — the widest such split Axis Intelligence Research has measured in a single industry to date.
- Axis Intelligence Research’s review of primary sources finds no support for the “$3.8 trillion in AI value” figure circulating across several finance-statistics aggregators. McKinsey’s actual published estimate for generative AI’s addressable value in banking is $200 billion to $340 billion annually; the $3.8 trillion figure instead describes 2025 revenue across the world’s 1,000 largest banks, an unrelated McKinsey data point.
- According to Axis Intelligence Research’s reading of the Evident AI Index, JPMorganChase ranks first overall among the world’s 50 largest banks on AI maturity, leading outright on three of the index’s four pillars — Innovation, Leadership, and Transparency — and finishing second only to Capital One on Talent.
- Per Axis Intelligence Research’s analysis of KPMG’s 2026 global survey, 71% of finance leaders report AI is meeting or exceeding ROI expectations, but only 23% say it is exceeding them outright, and firms that can efficiently produce AI audit evidence report three to six times the rate of measurable improvement of those that cannot.
- Axis Intelligence Research confirms the EU AI Act’s high-risk obligations for credit-scoring and insurance-pricing systems remain scheduled for enforcement on August 2, 2026 — unlike the biometrics, employment, and education-related high-risk categories that the EU’s 2025 “AI omnibus” agreement pushed back to December 2, 2027.
How Many Financial Institutions Use AI in 2026?
The honest answer depends entirely on which door you’re measuring from.
KPMG’s 2026 Global AI in Finance report surveyed 1,013 senior finance leaders across 20 countries and 13 sectors, all at companies with $250 million or more in annual revenue, fielded in March 2026. Among that population, more than three-quarters of organizations are already using AI in financial planning, reporting, and commercial analysis. Decision-making quality improved for 70% of respondents, decision-making speed for 71%, and forecasting accuracy for 64%. Organizations running agentic AI for finance separated from the rest of the pack by 32 percentage points on average — a gap that widens to nearly 40 points on forecast accuracy and ROI specifically.
That is a very different number from the one you get surveying finance leaders regardless of company size. A 2026 CFO Connect report, built on polling from General Atlantic, put AI adoption inside finance functions at 56% — described in the report as having doubled since 2023, but still the lowest of any business function inside the typical company. Within that 56%, only 17% of finance teams are using AI in core workflows; 45% remain in what the report calls “limited pilot” mode. Sixty-eight percent of CFOs say they’ve been slow to adopt because they don’t know where to start.
Neither number is wrong. They’re measuring different populations. KPMG’s respondents skew toward companies large enough to fund a dedicated AI program; CFO Connect’s sample includes the mid-market firms still deciding whether to run a pilot at all. Put the two side by side and a real story emerges: company size, not industry, may be the better predictor of whether a finance team has actually adopted AI.
The Back-Office AI Gap: Finance Departments vs. Frontline Risk Operations
Here’s the number that doesn’t show up in either report on its own. Set the CFO Connect finance-function figure (56%) against the fraud-detection adoption figure from Feedzai’s separate, industry-wide survey (90%) and you get a 34-point spread between how fast a bank’s own back office adopts AI and how fast that same industry’s frontline risk operations already have.
Axis Intelligence Research built this comparison — combining a finance-function survey and a financial-crime survey that were never published against each other — into a single figure: the Back-Office AI Gap Index (BOAG™). The math is simple by design: 90% minus 56% equals 34 points, or expressed as a ratio, finance-department adoption currently sits at just 62% of fraud-detection adoption in the same industry. Snapshot date: Q2 2026, based on the most recent available editions of both underlying surveys.
The gap is not really about technology readiness. Fraud detection has a business case that writes itself — every fraudulent transaction it catches is money the bank keeps. A finance department’s AI use case is fuzzier: better forecasts, faster closes, cleaner board decks. Harder to put a single dollar figure on, easier to defer. That is likely most of what’s driving the 34 points, and it also explains why the gap should close from the top down, as agentic AI tools give finance teams sharper, more measurable use cases to point budget at.
Which Banks Lead on AI? The Evident AI Index Rankings
If you want a single, defensible ranking of which banks actually run mature AI operations rather than which ones talk about AI the most, the Evident AI Index is the closest thing the industry has to a standard. Now in its fourth edition (October 2025), it benchmarks the 50 largest banks in North America, Europe, and Asia against 70+ public indicators across four weighted pillars: Talent (45%), Innovation (30%), Leadership (15%), and Transparency (10%). Evident has since extended the same methodology to insurance, payments, and — as of 2026 — a regional Middle East and Africa banking index, bringing its total coverage past 100 financial institutions.
| Rank | Bank | Talent | Innovation | Leadership | Transparency |
|---|---|---|---|---|---|
| 1 | JPMorganChase | 2 | 1 | 1 | 1 |
| 2 | Capital One | 1 | 2 | 20 | 17 |
| 3 | Royal Bank of Canada | 12 | 3 | 3 | 3 |
| 4 | CommBank | 4 | 13 | 4 | 2 |
| 5 | Morgan Stanley | 13 | 4 | 9 | 21 |
| 6 | Wells Fargo | 6 | 5 | 40 | 14 |
| 7 | UBS | 3 | 25 | 7 | 4 |
| 8 | HSBC | 14 | 8 | 15 | 6 |
| 9 | Goldman Sachs | 7 | 9 | 18 | 23 |
| 10 | Bank of America | 10 | 7 | 12 | 24 |
Source: Evident AI Index, 2025 edition (October 2025). Sub-pillar values are per-pillar rank out of 50, not scores.
JPMorganChase’s position is not an accident of one strong quarter. It has held the top spot on the index since Evident launched it in January 2023, and it is the only bank in the top 10 leading three of the four pillars outright. Capital One’s split ranking is the more interesting story in this table: first on Talent, but 20th on Leadership and 17th on Transparency — a bank with the engineering bench but not yet the public narrative or disclosure practices to match, at least by Evident’s methodology.
Sarah Davis, Axis Intelligence Research: I went looking for what “Leadership” actually measures before trusting a rank on it, because “AI narrative” sounds like exactly the kind of category a comms team can win without shipping anything. It isn’t just talk — Evident scores it against reported ROI and public use-case disclosure, not press-release adjectives. Fine. But it does mean a bank can outscore a rival on Leadership by disclosing more, not necessarily by doing more, and that’s a distinction worth holding onto every time this index gets cited as pure capability.
The insurance side of the index tells its own story: Allianz overtook AXA for the top spot in the 2026 ranking, and AI-specialist roles across the 30 tracked insurers grew 32% year-over-year even as those same insurers’ broader workforce shrank 2.2% — a headcount pattern that shows up nowhere in the adoption-percentage statistics most reports lead with.
How Much Are Financial Institutions Spending on AI?
JPMorganChase’s AI Investment
JPMorganChase discloses more than $18 billion in annual technology investment on its own corporate site — a figure the bank has held steady in its public messaging even as internal allocation has shifted hard toward AI. The bank’s own 2025 Investor Day materials, filed with the SEC, put 2025 firmwide technology expense at roughly $18 billion and disclosed granular AI adoption figures rarely seen outside a single institution: more than 200,000 users of the bank’s internal LLM Suite, roughly 100 generative AI solutions already in production, and more than 40,000 engineers using AI coding assistants. The same filing reported a 35% year-over-year increase in value generated from AI and machine learning across fraud prevention, credit decisioning, pricing, and marketing.
Reporting from American Banker on JPMorganChase’s February 2026 investor update put the bank’s total 2026 expense guidance at $105 billion, of which roughly $20 billion is earmarked for technology — a step up the bank’s leadership has tied directly to the competitive threat from payments and fintech rivals, not to AI spending in isolation.
Industry-Wide Value Potential — and a $3.8 Trillion Myth
McKinsey’s own published research puts generative AI’s addressable value for the banking industry at $200 billion to $340 billion annually, or 2.8% to 4.7% of industry revenue — driven mostly by productivity gains in risk reporting, regulatory monitoring, and data collection rather than entirely new revenue.
Sarah Davis, Axis Intelligence Research: That $200 billion to $340 billion range is the number McKinsey actually published, in June 2023, and it’s the number that still gets cited by name in the firm’s own banking research. So where does the $3.8 trillion figure — the one showing up across several “AI in finance” statistics roundups — actually come from? I traced it. It’s real, but it has nothing to do with AI. McKinsey’s 2026 Global Banking Annual Review reports that the world’s 1,000 largest banks generated a combined $3.8 trillion in revenue in 2025. Somewhere in the citation chain, a “$3.8 trillion” banking figure and a “McKinsey AI in banking” headline got stapled together by someone who didn’t read past the pull-quote. It’s a small error with a large multiplier — off by roughly 11 to 19 times the real estimate — and it’s exactly the kind of thing that happens when statistics pages compile from other statistics pages instead of the primary report.
Company-level disclosures back up the more conservative McKinsey range better than the inflated one does. JPMorganChase’s 35% year-over-year AI/ML value increase, applied against a bank with $173 billion in 2024 revenue, is a real but bounded number — nowhere close to trillion-dollar territory for a single institution, which is exactly what you’d expect if the industry-wide addressable value sits in the hundreds of billions rather than the trillions.
AI Fraud Detection Statistics in Finance
Feedzai’s 2025 AI Trends in Fraud and Financial Crime Prevention report surveyed 562 fraud and financial-crime professionals across retail banks, commercial banks, investment banks, credit unions, and fintechs in a 31-question survey fielded by an independent research firm in March and April 2025. Ninety percent of financial institutions now use AI to detect fraud, and two-thirds integrated it within the past two years — a rapid normalization for a technology that, in most of these institutions, was a pilot project five years ago.
Inside that 90%, AI is not evenly applied. Scam detection accounts for the largest single use case at 50% of institutions, ahead of transaction-fraud monitoring (39%) and anti-money-laundering screening (30%). Forty-three percent of financial professionals report measurably increased efficiency within their fraud teams as a direct result.
| Fraud-fighting AI use case | Share of financial institutions | Source |
|---|---|---|
| Scam detection | 50% | Feedzai |
| Transaction fraud monitoring | 39% | Feedzai |
| Anti-money-laundering (AML) screening | 30% | Feedzai |
Adoption has not been friction-free. Eighty-seven percent of banks cite data management — fragmented sources, inconsistent formats, regulatory constraints on what can be pooled — as their single biggest AI deployment hurdle, and that burden falls hardest on smaller institutions without the modernized data infrastructure JPMorganChase describes in its own investor materials. Eighty-nine percent of banks say they prioritize explainability and transparency in their fraud-detection AI specifically, a higher bar than most institutions apply to AI elsewhere in the business, for the obvious reason that a wrongly-flagged transaction has a customer on the other end of it.
How Criminals Are Using AI Against Financial Institutions
The same Feedzai survey found the other side of this equation just as stark: more than half of all fraud now involves AI in some form, and 92% of the financial-crime professionals surveyed say fraudsters are actively using generative AI against them. Only 8% say they see no GenAI-driven fraud at all.
The specific tactics break down unevenly. Sixty percent of respondents flag AI voice cloning as a major concern, 59% cite AI-powered phishing and SMS scams, 56% point to AI-driven social engineering more broadly, and 44% report deepfakes showing up directly inside fraud schemes — most often not to breach an account directly, but to build the trust a victim needs before handing money over voluntarily during a scam’s setup phase.
Sarah Davis, Axis Intelligence Research: “AI fraud detection” and “AI-powered fraud” are now running on the same underlying model families, sometimes the same open-weight releases, updated on roughly the same release cadence. That’s not a metaphor — it’s the actual arms-race structure the 92% figure describes. Banks are not defending against a static threat they can eventually solve for. They’re defending against an attacker with access to the same generation of tools, and the 87% citing data management as their top hurdle is really a story about who can retrain faster.
AI in Trading, Wealth Management, and Credit Underwriting
Algorithmic and AI-assisted systems have executed a majority of US equity trading volume for more than a decade — a market-structure fact well documented in SEC market-structure research and widely cited across the industry, though the specific percentage varies by source and by how “algorithmic” is defined, from execution-only order routing up through fully autonomous strategy generation. Axis Intelligence Research is treating any single precise figure here as directional rather than exact, pending a primary regulatory breakdown current enough to publish with confidence.
Robo-advisory and AI-assisted wealth management present a similar pattern: assets under management on robo-advisory platforms have grown substantially since the category’s inception in the early 2010s, but published market-size estimates for 2026 disagree by a wide margin depending on which research firm and which definition of “robo-advisor” is used — a gap Axis Intelligence Research will size precisely in a dedicated update once a primary, disclosed-methodology figure is available.
Credit underwriting is the one corner of this section with a hard regulatory line already drawn: the EU explicitly classifies AI systems used to evaluate creditworthiness as high-risk, with the narrow exception of AI used purely to detect financial fraud. That distinction — underwriting in, fraud detection out — is covered in full in the regulatory section below, and it is worth reading before assuming any AI credit tool is treated the same way under the law.
What Is the ROI on AI in Finance?
KPMG’s 2026 survey found 71% of finance leaders say AI is meeting or exceeding their return-on-investment expectations. Read no further and that sounds like a technology that has arrived. Read the next line and the picture narrows fast: only 23% say AI is exceeding expectations outright — a smaller, more demanding bar than “meeting or exceeding,” and the gap between those two figures is where most of the real story sits.
What separates the 23% from the rest isn’t the AI itself. It’s governance. Organizations that can efficiently produce audit evidence for their AI systems report error-reduction improvements at 33%, versus 6% for organizations that cannot — better than a five-to-one difference. On confidence in scaling AI further, the same governance-capable group reports improvement at 42%, against 14% for the rest — roughly three times higher. KPMG frames this as evidence that AI governance functions as an enabler of performance rather than a brake on it, which fits: the firms measuring their AI closely enough to produce clean audit trails are, unsurprisingly, also the firms that can tell whether the AI is actually working.
Thirty-six percent of surveyed finance leaders cite data quality, integration, and system interoperability as their single greatest opportunity to extract more value from AI — the same barrier Feedzai’s fraud-focused respondents named independently, in a different survey, of a different population, inside the same industry. When two unconnected surveys land on the same bottleneck without being asked to agree, that’s usually a real constraint rather than a coincidence of question design.
Is AI in Finance Regulated? The EU AI Act and Beyond
EU AI Act Timeline for Financial Services
The European Commission’s own AI Act guidance is unambiguous about where financial services sits inside the regulation: AI systems used to evaluate creditworthiness or establish a credit score are explicitly named as high-risk, with a single carve-out for AI used purely to detect financial fraud. Insurance risk pricing gets the same high-risk classification. Regulation (EU) 2024/1689 entered into force on August 1, 2024, and becomes fully applicable on August 2, 2026, with prohibited practices and AI-literacy obligations already in effect since February 2, 2025, and governance rules for general-purpose AI models in effect since August 2, 2025.
Here is the part most compliance guides currently miss, or get vague about on purpose: in May 2026, the EU reached political agreement on a “Digital Omnibus” package that delayed some high-risk obligations. But not all of them. The delayed categories — pushed to December 2, 2027 — are biometrics, critical infrastructure, education, employment, and migration/asylum/border control. Systems embedded in physical products such as lifts or toys were pushed further, to August 2, 2028. Credit scoring and insurance pricing are not on either delayed list. Barring a further amendment between now and then, the original August 2, 2026 date holds for the systems banks and insurers actually run day to day: credit decisioning, loan underwriting, and risk-based insurance pricing. Penalties for high-risk violations reach up to €15 million or 3% of global annual turnover; prohibited-practice violations reach up to €35 million or 7%.
The US Regulatory Landscape
US oversight has moved more incrementally and through existing supervisory channels rather than new legislation. Banking regulators, including the Office of the Comptroller of the Currency and the Federal Reserve, have begun asking banks to map their AI use across functions such as KYC and lending as part of ordinary supervisory examinations — a shift toward treating AI governance as a standard exam topic rather than a novel one. There is, as of this writing, no single US federal statute analogous to the EU AI Act; oversight instead runs through banking-specific model-risk-management guidance layered on top of general consumer-protection and fair-lending law.
Frequently Asked Questions
What percentage of financial institutions use AI in 2026?
It depends on which part of the institution you’re measuring. Ninety percent of financial institutions use AI for fraud detection, per Feedzai’s 2025 survey. Inside the finance department specifically, only 56% report using AI at all, per CFO Connect’s 2026 survey — though that rises to more than 76% among large firms with $250 million-plus in revenue, per KPMG.
What is the Back-Office AI Gap?
The Back-Office AI Gap Index (BOAG™) is an original Axis Intelligence Research calculation comparing finance departments’ own AI adoption (56%) against the same industry’s fraud-detection AI adoption (90%) — a 34-percentage-point gap, or a ratio of 62%, calculated by cross-referencing CFO Connect’s 2026 finance-function survey against Feedzai’s 2025 fraud-industry survey.
Which bank ranks first on the Evident AI Index?
JPMorganChase ranks first overall in the Evident AI Index’s 2025 edition, which benchmarks the world’s 50 largest banks across Talent, Innovation, Leadership, and Transparency. It has held the top position since the index launched in January 2023.
How much has JPMorganChase invested in AI?
JPMorganChase discloses more than $18 billion in annual technology investment on its own corporate site, and its 2025 SEC-filed Investor Day materials reported more than 200,000 users of its internal LLM Suite, roughly 100 generative AI solutions in production, and a 35% year-over-year increase in value generated from AI and machine learning.
Is the $3.8 trillion AI-in-finance figure real?
Not in the way it is usually cited. A $3.8 trillion figure circulates widely as “AI’s value to financial services,” but it does not appear in that form in McKinsey’s actual AI research. It is McKinsey’s separate estimate of 2025 revenue across the world’s 1,000 largest banks. McKinsey’s real, published estimate for generative AI’s addressable value in banking is $200 billion to $340 billion annually.
What percentage of financial fraud now involves AI?
More than half of all fraud now involves AI in some form, and 92% of surveyed fraud and financial-crime professionals report that fraudsters are actively using generative AI, according to Feedzai’s 2025 survey of 562 professionals.
When do EU AI Act rules apply to credit-scoring AI?
August 2, 2026. Credit-scoring and insurance-pricing AI systems are explicitly classified as high-risk under the EU AI Act and were not included among the categories the EU’s 2025 “AI omnibus” agreement delayed to December 2027.
What is the ROI on AI investments in finance?
Seventy-one percent of finance leaders report AI is meeting or exceeding ROI expectations, but only 23% say it is exceeding them outright, per KPMG’s 2026 survey of 1,013 finance leaders. Organizations with mature AI governance report three to six times the rate of measurable improvement of those without it.
Methodology
Axis Intelligence Research compiled this dataset from eight primary sources: KPMG’s 2026 Global AI in Finance report (1,013 respondents), Feedzai’s 2025 AI Trends in Fraud and Financial Crime Prevention report (562 respondents), the Evident AI Index (2025 edition, 50 banks, 70+ indicators), JPMorganChase’s own corporate disclosures and SEC-filed Investor Day materials, McKinsey’s published generative AI research, the European Commission’s official AI Act guidance, CFO Connect’s 2026 survey of finance leaders, and NVIDIA’s 2026 State of AI in Financial Services survey (800+ respondents). Each figure was traced to the issuing organization directly rather than to a secondary aggregator. Where a claim appeared across multiple statistics roundups but could not be traced to a primary, disclosed-methodology source — including specific percentages for algorithmic trading’s share of equity volume and current robo-advisory assets under management — Axis Intelligence Research presented the trend directionally rather than assert an unverified figure with false precision. The $3.8 trillion discrepancy documented above was identified by tracing a widely-cited figure back through its citation chain to two separate, unrelated McKinsey reports.
The Back-Office AI Gap Index (BOAG™) is an Axis Intelligence Research original metric, calculated as the percentage-point difference between financial institutions’ fraud-detection AI adoption rate (Feedzai, 2025) and finance departments’ own AI adoption rate (CFO Connect, 2026). Neither source publishes this comparison; Axis Intelligence Research constructed it by cross-referencing the two surveys’ closest comparable adoption metrics. Limitations: the two surveys were fielded roughly a year apart, use different sample sizes and methodologies, and measure adoption using each organization’s own definition of “AI use” — meaning BOAG should be read as a directional indicator of a real structural gap, not a single number with statistical precision to the decimal point.
About This Dataset
Dataset: AI in Finance Statistics 2026 |
Creator/Publisher: Axis Intelligence Research |
License: CC BY 4.0 — free to use, share, and adapt for any purpose, including commercial use, provided you credit Axis Intelligence Research and link to this page.
Update approach: revised when underlying data is superseded, a major regulatory or earnings event changes the picture, or ranking/traffic signals indicate the page needs refreshing — not on a fixed calendar.
Data collection window: figures reflect primary sources published or updated between March 2025 and June 2026; each is dated at first use above.
Citation
APA: Axis Intelligence Research & Davis, S. (2026, July 1). AI in finance statistics 2026: Adoption, spending, ROI, and fraud-detection data by institution. Axis Intelligence. https://axis-intelligence.com/ai-in-finance-statistics/
MLA: Axis Intelligence Research, and Sarah Davis. “AI in Finance Statistics 2026: Adoption, Spending, ROI, and Fraud-Detection Data by Institution.” Axis Intelligence, 1 July 2026, axis-intelligence.com/ai-in-finance-statistics/.
Chicago: Axis Intelligence Research and Sarah Davis. “AI in Finance Statistics 2026: Adoption, Spending, ROI, and Fraud-Detection Data by Institution.” Axis Intelligence. Last modified July 1, 2026. https://axis-intelligence.com/ai-in-finance-statistics/.
Related Axis Intelligence Research: AI Statistics 2026 for the broader market picture · AI in Business Statistics 2026 for the same adoption-reality-gap pattern across all industries, not just finance · Agentic AI Statistics 2025-2026 for the agentic-AI performance gap referenced above · AI Jobs 2026 for compensation and hiring data behind the Evident AI Index Talent pillar · AI Job Displacement Statistics for the workforce side of financial-services automation.