There are new warnings of a surge in the use of synthetic media to target the world’s insurers and banks.

Deepfake detection and synthetic media authentication, expert identifAI, has released new research examining how synthetic media is being used in financial fraud across banking, insurance and enterprise environments.

The findings highlight the growing role of AI-generated video, voice and identity manipulation, underscoring the urgent need for organisations to integrate real-time deepfake detection into existing fraud, identity, and verification controls.

The” Industrialisation of AI-Enabled Financial Fraud” tracks global incident telemetry through August 2026 and exposes how deepfakes have shifted from viral social content into a $40 billion financial fraud engine systematically weaponised against global banks, insurers, and corporate treasuries.

Criminal syndicates are deploying “Fraud-as-a-Service” toolsets and virtual camera-injection software to bypass legacy biometric verification, tricking enterprise workflows into authorizing massive financial losses.

Video and audio dominate financial fraud. Video synthesis is now the single largest fraud modality at 42.7%, closely followed by audio-based voice cloning at 21.9%. Combined, they represent 64.6% of all observed deepfake fraud.

The top geographies for deepfake frauds are major banking hubs. The United States leads as the primary target/source of deepfake fraud incidents at 34.1%, followed by India (7.7%) and the United Kingdom (6.6%).

Social media is used as the main reconnaissance ground. Bad actors heavily exploit X (42.1%) and LinkedIn (14.5%) to harvest high-fidelity video and audio of senior executives from podcasts, interviews, and public appearances to train generative AI models.

Bypassing Biometrics & KYC: Identity fraud explicitly accounts for 17.6% of incidents, with synthetic media now causing 1 in every 100 identity verification check failures globally. Text phishing is dead, high-fidelity audio/video impersonation is the new default.

What emerges from our findings is that while the first generation of deepfakes fooled our eyes, the next generation is engineered to fool our financial systems, specifically targeting banks and insurance companies,” said Marco Ramilli, co-founder & CEO at identifAI. “Deepfake detection can no longer exist as a post-incident forensic tool. Instead, research shows it needs to be embedded as an active, Zero-Trust security layer inside every critical financial workflow. There is a massive opportunity for financial institutions to act early and hedge this growing risk.”

High-profile breaches detailed in the report highlight the devastating financial impact of synthetic impersonation:

• The $25.6 million Arup Heist: A finance worker transferred HK200M (25.6M USD) across 15 transactions after attending a live video conference where every participant – including the Chief Financial Officer – was a real-time deepfake.
• ABN AMRO Bank: Attackers deployed ID morphing and deepfake liveness bypasses to open 46 fraudulent accounts before detection.
• Ferrari C-Suite Targeting: Attackers deployed a neural voice clone of the Ferrari CEO via WhatsApp audio notes to request an unauthorized currency-hedge transaction.

As autonomous AI agents, such as Meta Muse and autonomous execution bots gain mainstream momentum, identifAI’s warned that AI automation does not eliminate fraud – it dramatically amplifies the trust problem.

Digital workflows are shifting from the traditional pipeline (human to browser to bank) to autonomous loops (human to AI agent to browser to bank), making AI agents act as pre-filters for financial decisions on behalf of users. When a user instructs an agent to “Pay my bill,” the agent itself becomes vulnerable to manipulated inputs and synthetic disinformation. Securing next-generation finance requires a fundamental shift: it is no longer just about authenticating the human or the app, but equipping AI agents with a verification layer to accurately distinguish real media from malicious, synthetic inputs before executing transactions.

“Fraud used to need your password. Now it needs your face and a few seconds of your voice – and for most people, both are already online,” said Kate Burns, VP EMEA at identifAI. “This isn’t just a banking issue. The same attack works on a hospital, a council or a grandparent. The cost of faking a person convincingly has escalated, and our instinct to trust our own eyes hasn’t moved at all. The cost of synthetic fraud is already being passed down to consumers through higher banking fees and insurance premiums, as financial institutions grapple with these new forms of threats.”