Field guide
Synthetic Identity / AI-Generated Document Fraud
How it works
The mass-scale creation of fake identities using AI to generate passports, driver’s licenses, selfies, and supporting documents that bypass KYC and AML controls. 404 Media reported that OnlyFake advertised document images for $15 each. A later review of the service’s price list shows $15 for one document and lower unit prices in bulk. These pass basic OCR and human review. Combined with mule accounts and prepaid instruments, they create fully operational fake personas.
On the record
Sumsub, a verification company, said synthetic identity-document fraud in Europe rose 378% from the first quarter of 2024 to the first quarter of 2025. Sumsub described that fraud as documents generated with artificial intelligence. The figure is the company’s own count for that period. 404 Media reported that an ID image from OnlyFake passed the identity check at the crypto exchange OKX. The UK Gambling Commission said in 2025 that it is aware of more attempts to bypass customer due diligence with false documents, deepfake video, and face swaps made by artificial intelligence. A survey published by Sumsub said 78% of responding iGaming operators had seen more AI-generated fake documents in the past year.
Detection & preventionADVISORY
The paid halfHow a monitoring room catches synthetic identity / ai-generated document fraud, and what stops it recurring, written for the desk, not the reader.
Detection and prevention is the paid layer. It opens on all 400 methods with the downloadable edition, or with any paid plan.
Every subscription tier carries full access to the Cheats & Scams Library, Professional upward: the detection and prevention notes on all 400 methods, plus the searchable edition as a download.
Already subscribed? Sign in and this opens.
Published as a detection reference for surveillance, compliance and gaming-operations professionals. Thresholds and tuning are set by the property. Nothing here is instruction: the method is described so it can be recognised.