AI Utilised for Forging Art Documentation

AI aids in forging art provenance documents and certificates.

Artificial intelligence is aiding fraudsters in forging provenance documents and authenticity certificates for artworks, according to insurance broker Olivia Eccleston, as reported by the FT.

She noted that this trend has “added a new dimension to the age-old problem of forgery and fraud in the art market.”

An expert in loss adjustment within this sector revealed that he received dozens of appraisal certificates as part of an insurance claim for a collection of decorative paintings, which initially appeared convincing.

A claimant’s oversight led the analyst to suspect forgery. It turned out the documents were created using an automated system—the description field for each individual work was identical.

Other industry representatives highlighted that in some cases, artificial intelligence was used with malicious intent—fraudsters aimed to forge provenance documentation. There were also incidents where users asked neural networks to find information about their artworks, only for the AI to hallucinate.

Angelina Giovanni, co-founder of Flynn & Giovani, remarked that AI makes it easy to create fake documents due to its operational mechanics. A neural network is obliged to provide an answer, so if given enough information, “somewhere an LLM will guess.”

Experts emphasised that the use of artificial intelligence is a new twist on old practices. Whereas people once stole or forged forms with logos of reputable institutions, they now employ new technology.

“Throughout my career, I’ve seen many fake documents. This is not a new phenomenon, but AI has made it more realistic,” stated Filippo Guerrini-Maraldi, head of fine art at insurance company Howden.

Harry Smith, executive chairman of Gurr Johns, said that artificial intelligence has simplified and accelerated the work of fraudsters.

To combat this new phenomenon, digital document metadata can be checked for AI interference. This is the approach taken by Grace Best-Devereux, a loss adjustment specialist at Sedgwick.

Her colleagues do likewise—using LLM tools to verify document authenticity. However, this is becoming increasingly difficult due to the regular advancement of technology.

Earlier, the US Treasury Department utilised artificial intelligence to analyse data and recover over $4 billion obtained through fraud and misconduct in the 2024 fiscal year.