Of all the fraud trends worth watching into 2027, synthetic identity is the one rising fastest off its own base. Reports through 2026 describe a multiple-fold year-on-year increase, with synthetic identities now behind a meaningful and growing share of all fraud — and in some regions close to half of it. Generative AI is the accelerant: it makes a fabricated person look consistent, documented and real at a scale and polish that manual fraud never achieved.
What a synthetic identity is
A synthetic identity is not a stolen person; it is a manufactured one. Fraudsters combine real fragments — a genuine but unused national ID number, a real address — with fabricated details to build a plausible individual who does not exist. That identity opens accounts, builds a credit and transaction history over months to look legitimate, and then cashes out. Because no real victim notices early (there is no real person to raise the alarm), synthetics often mature quietly before they strike.
Why identity checks alone keep failing
Traditional verification asks "is this a valid, consistent identity?" A well-built synthetic is designed to answer yes. AI makes the documents cleaner, the backstories more coherent, and — paired with deepfake tooling — even the selfie and video checks passable. If your entire defence is to verify the identity the applicant presents, you are testing exactly the thing the fraudster engineered to pass. (We go deeper on the biometric angle in deepfakes beating KYC.)
The backstop: signals the identity can't control
The way to catch a convincing synthetic is to check things the fabricated identity does not control — above all, the network it operates from. A single human identity should map to a sane, consistent origin. Synthetic-identity operations do not:
- Many "people", one network. A farm of synthetic identities created and operated from the same hosting range, VPN or proxy pool is a pattern no set of unrelated real people produces.
- Origin that contradicts the story. An identity claiming one country, operating through anonymising infrastructure or an unrelated network, is worth a hard second look.
- Funding that doesn't line up. A card or bank whose issuing country, read from the BIN, does not match the identity's claimed location.
Corroborate identity with origin
You will not out-verify a synthetic on identity alone; you catch it by corroborating the identity against where it actually comes from. Fraudex supplies that independent signal — the network and type behind any IP, with a bot verdict, plus BIN intelligence — in one call, so a flawless-looking identity on a proxy farm still trips a wire. See the proxy detection API, the VPN detection API and the BIN lookup API.