Stay informed with free updates
Just sign up for artificial intelligence myFT Digest — delivered straight to your inbox.
Collectors are using documents generated by artificial intelligence to “prove” the authenticity and ownership of art when obtaining appraisals or making insurance claims, industry insiders say.
“Chatbot and [large language models] “We are helping fraudsters convincingly falsify sales invoices, valuations, provenance documents and certificates of authenticity,” said Olivia Eccleston, an art insurance broker at Marsh.
The trend “adds a new dimension to the long-standing problem of forgery and fraud in the art market,” she said.
An art loss adjuster reviewing claims on behalf of an insurance company was sent dozens of appraisal certificates as part of an insurance claim for a collection of decorative paintings.
Initially, the certificates seemed convincing, but the appraiser told the Financial Times that the descriptive fields for each work were the same. Complainant's apparent oversight led the expert witness to suspect that the certificate had been prepared by an automated writing system.
Industry officials said some of the AI uses were malicious, such as intentionally trying to falsify provenance documents. In another example, when people asked an AI model to search historical databases for references to their artwork, the results were “hallucinatory.”
Provenance is a term used in the art market to indicate the chain of ownership of an art object.
Angelina Giovani, co-founder of Flynn & Giovani, which researches the provenance of art, said it would be easy to use AI to generate false results. That's because AI is “very cunning… because it has to come up with an answer, and if you give it enough information, it will infer something.”
Giovani, who is developing an app to facilitate global provenance research, said he saw a document attached to the painting that clearly showed that AI had been used to sign the painting.
Experts say this use of AI is a new twist on an old practice.
Filippo Guerrini-Maraldi, head of art at insurance company Howden, said that while they used to steal or copy the letterhead of trusted institutions to show authenticity, they now use AI.
“I've seen so many forged documents in my career. That's not necessarily new, but AI has made it more realistic,” he said.
Art forgers have traditionally forged not only paintings, but also important documents that accompany them. Without this documentation, the work may be considered worthless.
Giovani said he saw fake ledger numbers and forged Nazi stamps on release documents.
Wolfgang Beltracchi, who painted hundreds of works, including works attributed to Max Ernst and Fernand Léger, also forged photographs to establish the provenance of his works.
“AI is making things that have been going on for a long time a little bit easier and a little bit faster,” said Harry Smith, executive chairman of art appraisal firm Gar Johns.
Grace Best-Devereux, an art loss adjuster at insurance claims firm Sedgwick, told the FT that she checks the metadata of documents submitted as digital files for clues about AI intervention.
Loss adjusters are also using AI tools to help authenticate source documents, she said. But recent advances in AI have made it more difficult for even experts to spot fraud, Best-Devereux said. “We're on this precipice, and it may be impossible to look at it and say, 'Looks like the text is wrong. We need to investigate further.'”
