Italian Police Used AI to Identify $200 M. Worth of Stolen Art and Antiquities

In the fall of 2024, Italian police used AI to analyze photographs taken at a popular antique fair, identifying a particular oil painting as having been one stolen from a church near L’Aquila 20 years prior.

The Carabinieri Command for the Protection of Cultural Heritage, Italy’s art crime squad, has been using an AI algorithm to identify stolen artworks and artifacts since 2023. The platform they use is called SWOADS—or Stolen Works of Art Detection System—and was developed by Salvatore Rapicavoli, an IT engineer who joined the art crime team in 2023.

A BBC article detailed the process, which began with Italian police officers visiting Mercanteinfiera, a 40-year-old “antiques village” that attracts exhibitors from all over Europe, in October 2024. When the officers uploaded their photographs into the art crime database, Swoads flagged a particular image of a young woman holding a scroll. The oil painting was a close—though, to the human eye, not an exact—match to a 17th-century painting of Saint Agnese that had been taken from Virgo Potens, a small church in Sulmona, in the Abruzzo region. At some point over the past two decades, the lamb Sant Agnese was cradling in the original had been painted over with a scroll inscribed with Latin text. Swoads, however, was able to identify “hidden visual fingerprints in the works—pairing similarities between shades of colour and rough silhouettes.”

Rapicavoli’s goal of using technology to solve art crime problems has been a success. Swoads has recovered more than 300 large paintings and statues, and more than 165,000 small objects like coins, letters, and medals, according to the BBC. All told, these objects are worth over $200 million. As for the portrait of Saint Agnese, it was returned to the church from where it was stolen. The thieves have yet to be apprehended.