Fraudsters have begun actively using neural networks to generate fake art authenticity documents, as well as invoices, the Financial Times (FT) reported, citing experts.
According to experts, collectors are using documents generated by artificial intelligence (AI) for such “confirmations“ about works of art when they appraise them or file insurance claims. “Chatbots and large language models are helping fraudsters convincingly forge invoices, documents confirming provenance and certificates of authenticity“, said Olivia Eccleston, art insurance broker at Marsh. She said the trend “has added a new dimension to the long-standing problem of forgery and fraud in the art market”.
A claims adjuster for insurers reported receiving dozens of appraisal certificates in connection with a claim involving a collection of decorative paintings. While they initially seemed convincing, on closer inspection, it turned out that “the description field for each individual piece was identical“. An apparent error on the part of the claimant led the expert to suspect that the certificates were created by an automated data entry system.
“Whereas before people stole or forged forms from reputable institutions to confirm authenticity, now they use artificial intelligence“, said Filippo Guerini-Maraldi, head of the fine arts department at Howden. “I have seen a lot of fake documents in my career, but artificial intelligence makes them look more realistic,“ he added.
Grace Best-Devereux, an expert at claims settlement firm Sedgwick, said she checks the metadata in digital documents for the use of artificial intelligence. However, she acknowledged that recent improvements in the quality of neural networks are making it harder for even experts to detect fraud.
According to Harry Smith, head of art valuation firm Gurr Johns, “there's no need to invent some kind of expert professor - you can just delegate that to artificial intelligence".