What fraud trends are defining 2026? (Featuring Alloy)

Alloy co-founder and President Laura Spiekerman has spent the last decade building one of the most widely adopted identity and fraud prevention platforms in financial services. With more than 700 banks, credit unions, and fintechs using Alloy, her vantage point into the fraud landscape is both broad and operational.

In this episode of Good Question, we took a different approach. Both Inscribe and Alloy recently published annual fraud reports: Inscribe focused on document fraud, Alloy on the broader fraud ecosystem. This conversation explored where those findings intersect, where they diverge, and what they reveal about the fraud trends actively shaping 2026.

The conclusion is clear: AI is accelerating fraud volume, blurring traditional categories, and forcing fraud teams to rethink where (and how) they apply controls. Just as importantly, fraud prevention is no longer viewed purely as a defensive function. In 2026, it is increasingly a growth lever.

Alloy’s view from the center of the identity stack

Laura reflected on Alloy’s origins in 2015, when identity was already complex—but far less adversarial than it is today. From the beginning, the stakes were clear: get identity wrong, and the consequences are severe. Get it right, and institutions unlock higher conversion, better customer experience, and stronger regulatory confidence.

What began as improved access to identity data has evolved into a platform spanning:

• 250+ data sources and partners
• configurable rules and decisioning logic
• auditability and regulatory explainability
• orchestration and analytics to adapt as fraud changes

That scale also underpins Alloy’s fraud report, which captures how fraud leaders across institutions are experiencing change in real time—not hypothetically.

One of 2026’s defining shifts: fraud is detected later in the lifecycle

One of the most striking trends in Alloy’s data is a decline in fraud detected at onboarding, paired with a rise in fraud detected post-transaction.

Laura framed this as both progress and risk:

• On the positive side, fraud teams are no longer treating onboarding as the only line of defense.
• On the risk side, onboarding fraud is becoming materially harder to detect as scams, money mules, and blurred intent complicate classification.

In 2026, the old buckets — first-party fraud, third-party fraud, scams — are no longer clean. AI-generated identities and assisted fraud make malicious behavior look increasingly legitimate, even to experienced teams.

Fraud prevention is now explicitly tied to growth

Another defining trend in 2026 is how fraud leaders frame success. More teams are recognizing that fraud strategy directly impacts growth.

Laura emphasized the shift:

• Reduce friction for legitimate customers
• Apply controls dynamically, not uniformly
• Avoid pushing good users into call center loops or manual reviews

AI is increasing fraud volume faster than sophistication

A headline stat from Alloy’s report: 91% of respondents observed more financial crimes committed using AI technology.

Laura made an important distinction. While AI is certainly increasing sophistication, the more immediate impact in 2026 is volume. AI enables fraudsters to generate attempts at scale, creating operational noise even when the underlying attacks are not advanced.

That volume creates pressure to “raise the gates.” But institutions that default to blanket friction risk undermining acquisition and retention. The winners in 2026 are those that can triage intelligently—letting good customers through while escalating uncertainty with precision.

Synthetic identity fraud is rising, and harder to define

Synthetic identity risk continues to grow, but Laura cautioned that the category itself is evolving:

• Some fraud that appears first-party is actually synthetic
  
• Many identities are hybrid, mixing real and fabricated attributes
  
• AI accelerates creation, testing, and iteration

The result is less clarity, not more. Classification, attribution, and downstream remediation all become harder, making layered defenses essential.

Ronan added that even when LLMs are not generating identities outright, they often act as fraud copilots, guiding bad actors through what to change, what to submit, and how to appear credible across documents, websites, and applications.

AI-driven fraud prevention only works if it’s actionable

Another defining signal for 2026: 82% of organizations are increasing investment in AI-driven fraud prevention.

Laura was careful to distinguish hype from reality. Machine learning models have existed in fraud for years. What matters now is whether institutions can:

1. Unify signals into a coherent decision system: Disconnected scores without orchestration, testing, and auditability don’t produce confidence.
   
   
2. Respond dynamically without shutting down the business: Detection alone is insufficient. Institutions need immediate fallback flows, like step-up verification, that preserve access while managing risk.
   

Her analogy was simple: an alarm system that rings but never triggers a response does not protect you.

Agentic AI is one of the most important fraud trends of 2026

When asked to name the fraud trend defining 2026, Laura pointed to agentic AI.

Not as a replacement for analysts, but as a way to eliminate the heavy operational burden of high-volume, low-risk work:

• sanctions and watchlist reviews
  
• PEP checks
  
• KYB verification steps
  
• repetitive false positives
  

These tasks consume enormous analyst time while rarely producing meaningful risk findings. Agentic systems can handle this work consistently, allowing humans to focus on complex, high-impact cases.

Ronan noted that many teams don’t even count spreadsheets and Google searches as “tools,” despite living in them daily. That’s where agentic systems can create immediate leverage.

Credit unions are experiencing sharp fraud increases for structural reasons

Alloy’s data shows credit unions reporting some of the largest increases in fraud events.

Laura offered several explanations:

• smaller budgets and fewer layers of protection
  
• heavier investment in digital channels, where fraud is most visible
  
• growing omnichannel experiences that outpace omnichannel fraud controls
  

Her warning for 2026: fraud flows toward gaps. As digital defenses improve, institutions must ensure branches, call centers, and cross-channel workflows are equally protected.

Ronan added that fraudsters optimize for efficiency. Institutions that lag in adoption may see disproportionate targeting, even as the overall fraud rate declines elsewhere.

The question fraud leaders must ask in 2026

Laura closed with a critical reframing:

Fraud teams measure how much fraud they stop… but they must also measure how much good business they block.

Declines, unnecessary friction, and abandoned applications represent real cost. A system with zero fraud but no growth is not a success.

In 2026, fraud leaders must understand both sides of the ledger.

The shared takeaway: layered, lifecycle-wide fraud strategy is no longer optional

If readers review both the Inscribe and Alloy reports, the message is consistent:

AI-driven fraud is increasing volume, distraction, and ambiguity. The response cannot be single-threaded.

Effective fraud strategy in 2026 requires:

• coverage across the full customer lifecycle
• multiple, complementary signal types
• governance that supports explainability and compliance
• automation paired with human judgment

Fraud teams are not being replaced. They’re being retooled.

Want to dive deeper into the fraud reports from Inscribe and Alloy? Check them out below: 

• Alloy’s 2026 State of Fraud Report 
• Inscribe’s 2026 State of Document Fraud Report