AI Is a Godsend for Criminals Forging Fake Art

For hundreds of years, humans have engaged in the fine craft of art forgery. Indeed, in 1496, a 21-year-old Michelangelo faked a piece of a Roman sculpture, passing it to the Roman antiquarian Cardinal Raffaele Riario to make a quick buck.

To pull off the con, the young Florentine artist rubbed acidic loam — dirt, basically — onto his own sculpture fragment in order to pass it off as one of the ancients. Had he been born a few hundred years later, he might have used ChatGPT.

New reporting by the Financial Times details the recent rise in art forgeries fueled by generative AI tools. Specifically, the outlet reports that art owners are using documents generated by AI in order to “prove” the authenticity and provenance of various pieces, to the frustration of fine art underwriters and brokers.

In the art world, provenance is the record of ownership of a specific piece, allowing collectors and art brokers to trace a work’s lineage.

“Chatbots and large language models [LLMs], are helping fraudsters convincingly forge sales invoices, valuations, provenance documents and certificates of authenticity,” Olivia Eccleston, a fine art broker at insurance firm Marsh McClennan, told the FT.

According to Eccleston, AI has “added a new dimension to an age-old problem of fakes and fraud in the art market.”

One insurance adjuster who spoke to the publication says they were sent dozens of certificates as part of a loss claim on a large collection of paintings. Though the documents “seemed convincing,” clues in the metadata led the adjuster to conclude that the entire art collection was a fake.

Interestingly, the FT notes that only a portion of documented AI use for fraud was intentional. In other cases, collectors used AI to find provenance of various works in reference databases — results that the tech hallucinated.

One analyst at art research firm Flynn and Giovani called AI in that context “quite conniving,” saying it “has to come up with an answer, so if you give it enough information, it will guess something.”

While adjusters are also utilizing AI to help keep up with the rising tide of fakes, Grace Best-Devereux, a claims adjuster at the claims investigation firm Sedgwick, told the FT it’s becoming difficult to stay ahead of the curb, given the ease with which fraudsters can now fake highly-probable documents.

“We’re at this precipice where it might not be possible for me to look at it and say ‘the text looks wrong, and I need to investigate this further,’” she said.

More on art: Acclaimed Movie Secretly Contained AI Generated Imagery