Sumsub Identity Fraud Report: Deepfakes Surge 31% as AI Agents Become New Attack Vector
By Effie Foxtrot · 17 November 2026
In brief: Sumsub's annual report finds that while overall identity fraud fell to 1.6%, deepfake attacks rose 31% and 53% of businesses now face fraud involving AI agents that can autonomously move money.
| Global Fraud Rate | 1.6% |
|---|---|
| Deepfake Fraud Increase | 31% YoY |
| AI Agent Fraud Exposure | 53% of businesses |
| FinServ Fraud Rate | 65% saw agent-related fraud |

Global identity fraud rates have dropped to 1.6% this year, a significant decline from the 2.6% peak recorded in 2024. However, this volume decrease masks a surge in sophistication, as criminals increasingly target the AI agents integrated into modern banking and checkout systems, making attacks significantly harder for traditional security frameworks to detect.
What was announced
The findings, detailed in the latest annual Identity Fraud Report, reveal a fundamental shift in the threat landscape. While the overall frequency of attacks has diminished, the new Fraud Complexity Index has risen across all six tracked global regions. This suggests that while basic "spray and pray" fraud attempts are failing, high-level, technologically advanced attacks are becoming more prevalent and difficult to mitigate.
A primary driver of this complexity is the weaponization of artificial intelligence. Deepfake fraud has increased by 31% year-over-year, even as the broader fraud rate fell. The report highlights a growing vulnerability in automated systems: 53% of businesses reported detecting or suspecting fraud involving an AI agent within the last year. These incidents are not merely theoretical; in 30% of these cases, the AI agent successfully moved money. Furthermore, 49% of these agent-related incidents involved a legitimate AI agent being manipulated through malicious instructions rather than a purely fraudulent bot.
The financial services sector remains the primary target, with 65% of firms reporting agent-related fraud. Despite this high exposure, corporate defenses appear to be lagging. Only 37% of businesses currently verify who has authorized an AI agent to act on their behalf, and only one in eight organizations maintains the capability to "switch off" an agent once it has been deployed into their ecosystem.
The companies involved
Sumsub is a global verification platform providing integrated tools for Know Your Customer (KYC), Know Your Business (KYB), and Anti-Money Laundering (AML) compliance. The company focuses on the entire user lifecycle, offering solutions that orchestrate verification processes to prevent identity theft and financial crime. By monitoring the "Fraud Complexity Index," the firm tracks how regional threat actors adapt to new security measures, providing data-driven insights into the evolution of digital deception.
Operating in a market increasingly defined by the tension between user friction and security, the company has positioned itself as a specialist in detecting AI-generated synthetic identities. Its platform is designed to handle the transition from human-centric verification to the emerging "agentic" economy, where non-human entities perform transactions. The firm maintains a significant presence in the fintech news cycle, with extensive reporting on its partnerships and technical integrations across the global financial landscape.
What FF News has reported before
FF News has tracked the steady expansion of Sumsub’s technological capabilities, particularly its focus on automated and on-chain security. Recently, the company launched its KYA Module to Prevent Unauthorized AI Agent Actions, a direct response to the vulnerabilities highlighted in its latest report. In the maritime sector, the company partnered with MarCoPay to protect Filipino seafarers from AI-enabled fraud. Additionally, Sumsub has worked to streamline business onboarding, as seen when it accelerated global venue onboarding for Playtomic using AI-powered KYB tools. The firm also joined the Fireblocks-Led OTL Initiative to help standardize on-chain identity and compliance standards.
What this means
The decline in total fraud volume is a false dawn for the fintech industry. The data suggests that the "easy" fraud is being caught by basic automation, forcing criminals to adopt "agentic" strategies that exploit the very AI tools banks use for efficiency. This creates a new category of risk: programmable liability. If an AI agent can be manipulated into moving funds, the traditional concept of "authorization" is broken. The industry is now under immense pressure to develop "Know Your Agent" (KYA) protocols. Without the ability to verify the intent behind an automated instruction, the financial sector remains highly vulnerable to sophisticated, high-value theft.
Frequently asked questions
How much has deepfake fraud increased according to Sumsub?
Deepfake fraud rose by 31% year-over-year, according to the Sumsub Identity Fraud Report. This increase occurred even as the overall global identity fraud rate fell to 1.6%, suggesting that fraudsters are shifting toward more sophisticated, AI-driven methods to bypass security.
What is the risk of AI agents in banking and finance?
The primary risk is unauthorized financial movement; 30% of businesses reporting AI agent fraud saw the agent move money. Additionally, 49% of incidents involved a legitimate agent being manipulated by malicious instructions, yet only 1 in 8 businesses has the ability to switch a compromised agent off.
Which industry is most affected by AI agent fraud?
Financial services are the most targeted sector, with 65% of businesses in this industry reporting incidents of agent-related fraud. This is significantly higher than the general business average, highlighting the vulnerability of automated banking and checkout systems.
More Fraud & Digital Identity news →
About the companies
Companies in this story: Sumsub