Fraudsters are using AI systems to forge art documents

Art galleries and insurers are dealing with a flood of forged documents built with AI, a trend now changing how people file claims, seek valuations, and defend ownership records, according to the Financial Times.

One fine art loss adjuster allegedly said they received dozens of valuation certificates for decorative paintings submitted in one claim. Each sheet looked correct at first, but the descriptions for different pieces were identical.

That single detail pushed the adjuster to suspect the entire batch came from an automated system. That check opened the door to wider concerns about how often forged files are slipping into the process without anyone noticing.

Exposing how fraudsters build convincing paperwork

Olivia Eccleston, a fine art insurance broker at Marsh, said chatbots and large language models are now helping fraudsters forge sales invoices, valuations, provenance documents, and certificates of authenticity.

Olivia said this adds a new layer to an old fraud problem in the art market. Some attempts are deliberate. Others start when someone asks an AI model to search historic databases, and it produces results that never existed. Those errors then show up in paperwork that gets sent to insurers as if it were fact.

The chain of ownership, known as provenance, is central in the art world. When people corrupt that chain with invented details, the artwork’s value collapses.

Angelina Giovani, co-founder of provenance research group Flynn & Giovani, said AI makes this easy because “it’s quite conniving… it has to come up with an answer, so if you give it enough information, it will guess something.” Angelina said she saw a case where an AI system appeared to create a signature on a painting to strengthen its story.

Experts note that none of this is new in principle. People once copied letterheads from respected institutions or designed fake stamps.

Now they lean on AI to generate the same paperwork with smoother language and fewer obvious errors. Filippo Guerrini-Maraldi, head of fine art at insurer Howden, said he has seen many forged documents over his career and that automated systems now make them look more realistic.

Angelina said she has seen fake ledger numbers and forged Nazi-era stamps on provenance files. She also pointed to the case of Wolfgang Beltracchi, who created hundreds of paintings and used staged photographs to build fake ownership histories behind them. These tactics show how far people go to support artwork that cannot stand on its own.

Tracking digital clues as fraud becomes harder to see

Harry Smith, executive chair of art valuers Gurr Johns, said AI now makes fraud quicker because people no longer need to invent a fake expert to back their claim. The tool produces whatever support text they want.

Grace Best-Devereux, a fine art loss adjuster at Sedgwick, said she checks metadata in digital documents to spot signs of AI interference. Grace said adjusters also use their own systems to decide whether a provenance document is real.

But she warned that the job is getting harder because new tools are making forged text look normal. Grace 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.’”

If you're reading this, you’re already ahead. Stay there with our newsletter.