AI fraud has changed the fraud threat landscape. While modern fraud targets everyone, small and medium companies are affected the most.
In this report, experts from Socure, Accenture, IBM and other companies talk about the current fraud landscape, how criminals use AI, who is being targeted and what anti-fraud tech and strategies are the most effective for businesses of all sizes.
AI Fraud Surges by 800% and Impacts 7.4 out of 10 Businesses
Socure’s report found that AI-based fraud attacks surged more than 8,000% across all industries and companies of all sizes. AI is allowing fraudsters to scale their operations to never-seen-before levels. One fraud ring tracked by Socure executed 38,241 transactions across 465 organizations in 30 industries within a single 90-day window.
“Before AI, a face, document, or voice could be trusted as real, but today AI lets anyone fabricate identities, clone voices and launch cyberattacks at a speed and scale no business can handle alone,” Matt Thompson, president and chief commercial officer at Socure, the AI-powered identity verification platform, told me.
While traditional fraud operations used to focus mostly on large enterprises, today AI-enabled fraud impacts businesses of all sizes, said Thompson.
The 2026 AFP Payments Fraud Survey found that 76% of organizations saw fraud attempts or losses to fraud attacks in 2025. The report highlights the increasing trend of impersonation tactics and AI-enabled fraud, including deepfake technologies.
“Large organizations usually have treasury teams, bank relationships and insurance to help recover losses,” Karthik Karunanithi, solution architect, IBM, told me. ”Smaller businesses often absorb the full impact,” said Karunanithi.
“Fraud now runs as a subscription business,” Nic Adams, cofounder and CEO of 0rcus, an AI-native offensive cyber company, told me.
“Stolen credentials cost a few dollars, a convincing CEO voice clone takes one podcast clip and generative AI writes thousands of personalized attacks before lunch,” said Adams.
“Fraud has matured into a full operational intrusion; attackers work your employees, your payment rails, your executive decision-making the way a con artist swindles a room,” Adams said.
Who Do AI Fraud Cybercriminals Target?
In its Redefining Cyber Resilience 2026 report, Accenture notes that AI-enabled cyberattacks increased 89% year over year in 2025. Besides embracing new tech, fraudsters are expanding their target groups and the fraud attack chain now includes supply chains, partners and customers.
“We’re seeing them target people such as employees with access to systems, executives with decision-making authority, suppliers and partners involved in payments and customers whose identities or accounts can be exploited,” Harpreet Sidhu, global cybersecurity lead at Accenture, told me.
“Anyone with wire authority has a giant target on their back, but fast-growing mid-market companies are getting mauled right now,” Jonathan Mitchell, financial industry lead at Founder Shield, a risk management partner, told me.
“Why? Because their revenue is exploding, but their security controls are often still stuck in ‘garage startup’ mode — and criminals excel at exploiting that exact gap,” said Mitchel.
From Healthcare to Military Families to AI Agents: Prime AI Fraud Targets
While, as mentioned, modern fraud affects everyone, specific industries like healthcare face unique challenges. The average cost of a breach in the healthcare industry in the U.S. is staggering and set at $10.22 million per breach, according to IBM. Fraudsters and cybercriminals use the data from these breaches to launch large-volume operations.
“Healthcare organizations remain a particularly valuable target because they manage some of the most sensitive information that exists,” Mike Hoey, CEO of Meridian, a healthcare and life sciences software engineering firm, told me.
But fraudsters are not just targeting businesses; specific communities are especially susceptible.
“Military families are a prime example of this,” said Thompson from Socure. Scammers exploit their steady government incomes and benefits, the disruption of frequent relocations and a tight-knit culture built on shared trust, Thompson said.
Other areas where fraud is expanding include gaming, government, crypto, marketplaces, big tech, AI platforms, minors and even AI agents. As AI agents are handed the keys to access systems, approve requests and even make purchases, they become valuable targets, said Karunanithi from IBM.
“If an attacker compromises an AI agent, they’re effectively gaining the same access as a trusted employee, except the agent operates at machine speed,” Karunanithi explained.
Cloud Security Alliance research found that only 28% of organizations can trace an AI agent’s actions back to the person who authorized them. “Most organizations can’t tell when that (AI Agent fraud) happens,” Karunanithi said.
The Tech and Strategies Helping Businesses Reduce AI Fraud
The AI fraud landscape, its scale and volume, may seem too overwhelming for decision-makers to handle. However, businesses of all sizes today can adapt and respond to modern fraud by deploying strategies and more cost-efficient tools that level up their tech stacks.
For example, Socure recently announced Socure Launch, which provides a self-serve option that can help start-ups and smaller businesses access the same enterprise-tested turnkey fraud and identity workflows that the top banks in the U.S. use today.
“Enterprise-grade fraud protection and infrastructure used to be something only the biggest institutions could afford,” said Thompson.
Speaking about anti-fraud tech strategies, Thompson noted a layered defence and Zero Trust are non-negotiable.
“The only approach is a layered defense system that includes both passive signals, such as an identity fraud score and active checks when risky users are identified, such as document verification,” said Thompson.
“Best-in-class organizations are unifying these signals into a single platform rather than manually stitching together separate tools for identity checks, fraud detection and risk decisioning,” Thompson explained.
On the other hand, Thompson said, Zero Trust allows companies to move away from one-time checks and toward continuous identity verification. Sidhu from Accenture agreed. “The best fraud defense is a Zero Trust design”.
“Organizations need to constantly verify identities, monitor behavior in real time and train people to spot AI-powered deception,” said Sidhu. “Make impersonation, process manipulation and unauthorized payments harder to execute and easier to stop.”
Adams from 0rcus said that what actually reduces fraud are controls that interrupt fraud while it is still in motion. This includes independent payment verification, phishing-resistant authentication, behavioral analytics, device intelligence, transaction limits and hard separation of approval authority.
“Think of it as putting a bouncer at every door money walks through,” said Adams.
Red Team Simulations and AI Agent Fraud Security
Raising awareness among your teams is also important, with red team fraud simulations being specifically useful.
“Pressure-test your people against realistic attacks, because the first deepfake call your CFO hears should come from your own red team,” said Adam.
Speaking about AI agent anti-fraud, Karunanithi said businesses should treat agents the same way they treat employees. “Give each one its own identity, limit its permissions to only what it needs and keep a record of every important action,” said Karunanithi.
Limiting what tools and accounts an AI agent can access is vital. “Too many organizations still rely on shared credentials or give AI agents more access than they actually need,” said Karunanithi.
The Road Ahead for AI Fraud Defense
Modern AI fraud is today an industrial, transnational cybercriminal high-volume threat targeting all businesses and communities. Sidhu from Accenture said fraud prevention has to be treated as a business resilience issue, not just a cybersecurity or compliance function.
The companies that reduce fraud most effectively will be the ones that combine strong technology with clear accountability, employee awareness and a culture where people feel empowered to challenge suspicious activity, said Sidhu.
On the other hand, Karunanithi from IBM said he thinks most organizations are solving the wrong problem. “They spend a lot of time trying to detect fraud after it happens instead of making it harder for fraud to happen in the first place,” said Karunanithi.
While both technology and fraud will continue to evolve and change, anti-fraud fundamentals will remain the same, said Karunanithi.
“I spent a decade in the U.S. Army as a Special Operations Officer, with four combat deployments to Iraq and Afghanistan,” said Thompson. “The tactics fraudsters use remind me a lot of what I saw overseas: decentralized, networked, fast to adapt and often invisible until the damage is already done.”
To fight fraud, companies of all sizes and from all sectors and industries need to collaborate, Thompson noted. “Fraud is too advanced to solve in silos; we need to come together,” said Thompson from Socure.
