Stay informed with free updates
Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.
Collectors are using documents produced by artificial intelligence to “prove” artworks’ authenticity and ownership when obtaining valuations or making insurance claims, according to industry figures.
“Chatbots and LLMs [large language models] are helping fraudsters convincingly forge sales invoices, valuations, provenance documents and certificates of authenticity,” said Olivia Eccleston, a fine art insurance broker at Marsh.
The trend has “added a new dimension to an age-old problem of fakes and fraud in the art market”, she said.
One fine art loss adjuster, who reviews claims on behalf of insurers, was sent dozens of certificates of valuation as part of an insurance claim on a collection of decorative paintings.
While the certificates initially seemed convincing, the adjuster told the Financial Times that the description field for each distinct work was the same. The apparent oversight by the claimant led the adjuster to suspect that the certificates were produced with an automated writing system.
Industry figures said some of the AI use was malicious, with deliberate attempts to forge provenance documentation. Other examples occurred when people asked an AI model to find references to their artwork in historic databases and it had “hallucinated” the results.
Provenance is the term used in the art market to demonstrate the chain of ownership of an artwork.
Angelina Giovani, co-founder of art provenance researchers Flynn & Giovani, said it was easy to use AI to create fake results because it was “quite conniving . . . it has to come up with an answer, so if you give it enough information, it will guess something”.
Giovani, who is developing an app to encourage global provenance research, said she had seen a document accompanying a painting where AI had apparently been used to apply a signature to the picture.
Experts say such use of AI is a new twist on old practices.
Whereas previously people stole or mocked up letterheads from reputable institutions to indicate authenticity, now they were using AI, said Filippo Guerrini-Maraldi, head of fine art at insurer Howden.
“I have seen so many forged documents in my career. It’s not necessarily a new thing — but what AI has done is made it more realistic,” he said.
Art forgers have traditionally not only faked paintings but also the important documentation that goes with them. Works can be considered valueless without this paperwork.
Giovani said she had seen fake ledger numbers and forged Nazi stamps on provenance documents.
Wolfgang Beltracchi, who painted hundreds of works, including some attributed to Max Ernst and Fernand Léger, also faked photographs to create his works’ provenance.
Harry Smith, executive chair of art valuers Gurr Johns, said: “AI makes what has been going on for a long time a bit easier and a bit quicker. You don’t have to dream up some professor expert — you can just get AI to do it for you.”
Grace Best-Devereux, a fine art loss adjuster at claims specialist Sedgwick, told the FT she was checking metadata on digitally filed documents for clues about AI meddling.
Loss adjusters have also used AI tools to help authenticate provenance documents, she said. However, recent improvements in AI have made it more difficult even for specialists to spot frauds, Best-Devereux said. “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’.”
