Art galleries and insurance companies are dealing with a torrent of fraudulent documents created using artificial intelligence, a trend that is now changing how people file claims, search for appraisals, and defend property records, according to the Financial Times.
One fine art loss adjuster allegedly said they received dozens of certificates of appraisal for decorative paintings submitted in one claim. Each sheet looked correct at first, but the descriptions of the various pieces were identical.
This single detail led the officer to suspect that the entire payment came from an automated system. This examination has opened the door to broader concerns about how often fraudulent files creep into the process without anyone noticing.
Revealing how scammers build convincing cards
Chatbots and large language forms are now helping fraudsters falsify sales invoices, appraisals, provenance documents and certificates of authenticity, said Olivia Eccleston, a fine art insurance broker at Marsh.
This adds a new layer to the age-old problem of fraud in the art market, Olivia said. Some attempts are deliberate. Others start when someone asks the AI model to search historical databases, producing results that didn’t exist before. These errors then appear in paperwork that is sent to insurance companies as if they were true.
The chain of ownership, known as provenance, is central to the art world. When people spoil that series with invented details, the value of the artwork collapses.
AI makes this easy because it’s “quite conspiratorial…it has to come up with an answer, so if you give it enough information, it’ll guess something,” said Angelina Giovanni, co-founder of the Source research group Flynn & Giovanni. Angelina said she saw a case where the AI system appeared to create a signature on a painting to reinforce her story.
Experts point out that none of this is new in principle. People once copied letterheads from respected institutions or designed fake stamps.
Now they rely on artificial intelligence to create the same papers with smoother language and fewer obvious errors. Filippo Guerrini Maraldi, head of fine arts at insurance company Howden, said he had seen many forged documents during his career, and automated systems now make them look more realistic.
Angelina He said I’ve seen fake ledger numbers and forged Nazi-era stamps on source files. She also pointed to the case of Wolfgang Beltracci, who painted hundreds of paintings and used photographs to construct a false ownership history behind them. These tactics show the lengths people will go to support works of art that cannot stand on their own.
Track digital clues as fraud becomes harder to see
Harry Smith, chief executive of art grading company Gurr Johns, said AI now makes fraud faster because people no longer need to invent a fake expert to back up their claims. The tool produces any support text they want.
Grace Best-Devereux, a fine art loss adjuster at Sedgwick, said she checks metadata in digital documents to detect signs of AI interference. Officers also use their own systems to determine if a source document is authentic, Grace said.
But she warned that the task is becoming more difficult because new tools make forged text look natural. “We’re at that precipice where it may no longer be possible for me to look at it and say, ‘The script looks wrong, and I need to investigate this further,'” Grace said.
Register on Paybit And start trading with $30,050 in welcome gifts
