People familiar with the astronomical prices of today’s art market might not bat an eye. But when the London National Gallery purchased Peter Paul Rubens’ Samson and Delilah in 1980, its price tag of $5.4 million (around $18 million today) made headlines as the third-highest sum ever paid at auction for a work of art.
Critics have long claimed that the costly 17th-century work is a fake. Those concerns were renewed this week, when Swiss company Art Recognition announced that it had analyzed the painting with artificial intelligence (A.I.) technology.
All told, writes Dalya Alberge for the Observer, the algorithm reported a 91.78 percent probability that Samson and Delilah was painted by someone other than Rubens. Comparatively, when scientists applied their A.I. technology to another Rubens work in the National Gallery, A View of Het Steen in the Early Morning (1636), they obtained a probability of 98.76 percent in favor of the work’s authenticity. (Art Recognition claims to have studied 148 such “uncontested” Rubens in a similar manner.)
“The results are quite astonishing,” Carina Popovici, Art Recognition’s co-founder and the scientist who led the study, tells the Observer. “I was so shocked. … We repeated the experiments to be really sure that we were not making a mistake and the result was always the same. Every patch, every single square, came out as fake, with more than 90 percent probability.”
Standing more than six feet tall, Rubens’ composition depicts Old Testament hero Samson sprawled in the lap of his lover, Delilah, who has just betrayed the secret of the sleeping man’s supernatural strength. An accomplice following Delilah’s instructions cuts the strongman’s hair, rendering him powerless—and vulnerable to the guards waiting just outside the door.
According to ArtWatch U.K., scholars generally agree that Rubens painted a version of Samson and Delilah around 1609 or 1610 for his Antwerp friend and patron, a well-known government official named Nicolaas Rockox. This timeline is supported by a preparatory painting in the Cincinnati Art Museum’s collections, a contemporary engraving by Jacob Matham and a painting by Frans Francken the Younger titled Banquet at the House of Burgomaster Rockox (1630-35). In this interior view of Rockox’s richly decorated home, Samson and Delilah can be seen hanging above the mantelpiece.
After Rockox’s death in 1640, the biblical scene vanished from the historical record until 1929, when it—or a copy of it—resurfaced in Paris. Famed German scholar Ludwig Burchard identified the painting as a genuine Rubens, but vocal critics of the attribution—among them independent scholar Euphrosyne Doxiadis—argue that the work may have disappeared, only to be replaced by a fake, as Edward M. Gómez reported for Das Spiegel in 2005.
Over the decades, scholars have singled out a few key discrepancies to build their case that the Baroque Flemish masterpiece is a counterfeit. In the 17th-century depictions of Samson and Delilah mentioned earlier, Samson’s whole foot is included in the frame, appearing near the lower righthand side of the composition. In the National Gallery version, meanwhile, that same foot is truncated by the edge of the canvas.
If the new A.I. analysis reported by the Observer is confirmed, it could lend more legitimacy to critics’ claims. Popovici tells the Observer that the team’s algorithm operates by teaching A.I. to identify the patterns of a particular artist’s unique brushstrokes.
The analysis referenced by Popovici has not yet been published in full. Though some experts, including Michael Daley of ArtWatch U.K., who describes the research as “exceedingly damning,” are convinced by the initial findings, others remain skeptical. As British art historian Bendor Grosvenor writes on Twitter, “The only thing this tale should tell us is that computers still don’t understand how artists worked. And probably never will.”
Grosvenor directed his readers to a 1983 technical bulletin in which National Gallery conservator Joyce Plesters examines the painting’s history, paint composition, structure and more. In Grosvenor’s view, Plesters’ analysis “show[s] the picture is indeed by Rubens.”
A museum spokesperson told the Observer that the gallery “always takes note of new research.” They added, “We await its publication in full so that any evidence can be properly assessed.”