Deepfake Identity Fraud Projected to Rise Nearly Sixfold in 2026, Industry Report Finds

TL;DR — Shufti's Identity Fraud Index Report projects that deepfake-powered identity fraud will rise 495% in 2026 compared with 2025, close to a sixfold increase in a single year. The report, based on more than one million fraud attempts analysed across Shufti's global verification network, splits AI-driven fraud into four types: synthetic identities, live video deepfakes, face swaps and document deepfakes. Document deepfakes are the fastest-growing category at close to +3,900% year on year. The report's conclusion is that manual review can no longer reliably catch high-quality deepfakes and that single selfie checks are no longer sufficient as a control.

Read the full report: Deepfake-Powered Identity Fraud Is Surging in 2026: Shufti's Identity Fraud Index Report, Shufti, 2026.

The headline number

The report's central projection is a 495% increase in deepfake-powered identity fraud in 2026 versus 2025. Shufti describes it as the sharpest year-on-year acceleration in its dataset.

Four attack types

• Synthetic identity. 42.3% of AI-driven fraud attempts in 2025, with a projected +173% growth in 2026. Real and fabricated data are blended to create a person who does not exist.
• Live video deepfakes. 28.1% share. An established threat that has grown every year since 2023.
• Face swaps. 17.6% share, projected to grow +121% in 2026.
• Document deepfakes. 11.9% share but the fastest-growing type, at nearly +3,900% year on year, as generative AI makes convincing identity documents cheap to produce.

Why it matters

The report argues that the cost and skill required to mount a credible deepfake attack have collapsed, which moves the threat from a niche concern to a baseline assumption for any remote onboarding flow. It cites Deloitte's estimate that generative AI could enable up to US$40 billion in fraud losses in the United States by 2027.

Shufti's recommended response is layered: capture integrity checks, liveness verification and forensic analysis working together, rather than relying on a single selfie or document check. The findings sit alongside similar warnings this year from the World Economic Forum and Sumsub on AI-generated fake IDs and their effect on KYC controls.