Walk into the basement of John Zemanovich home nestled in the small town of Pelham, Ont. and the entire back wall and part of the floor will be covered with beautiful paintings, filled with rich colours, purposefully placed lines and intricate designs. You’d be forgiven for believing paintings of such beauty should be displayed prominently above ground somewhere, perhaps in a museum or at least over a mantle. In fact, at one point, these paintings were making their way around Canada’s art scene for as much as $700,000. But now, they are collecting dust in a basement because they were recently discovered to be fake. “The (owners) bought them thinking they were real and they found out that they’re not real,” said Zemanovich, who works at an art authentication firm called Art Experts Canada. “These are a group of paintings that we have taken off the hands of previous owners when we were asked for our opinions on authentication. We have an authentication program that if somebody finds out their painting is fake, we offer them the ability to turn the painting back over to us and get a significant refund on our fees,” said Zemanovich. His firm has removed more than 50 forgeries related to one artist off the art market so far this year. The paintings were being circulated as authentic pieces by one of Canada’s most famous artists, Norval Morrisseau. Ever since Morrisseau, who is deemed the “Picasso of the North,” died nearly two decades ago, thousands of fake copycat paintings have flooded the market. Zemanovich and his team at Art Experts Canada have compared the fakes with hundreds of Morisseau’s real paintings, to determine that some of the styles used in the forgeries did not match his painting patterns. “We maintain a database that includes all the information about what kind of art he was producing, what kind of colours he’d be using at any given time, (and) where he was at different times in his career,” said Jonathan Sommer, CEO of Art Experts Canada, who believes there is still anywhere between 4,000 to 6,000 fake Morrisseau paintings still in circulation. “Once (you) accumulate a huge amount of data like that, and you couple that with looking at thousands of authentic and fake paintings, you start to develop what’s known as ‘connoisseurship,’ which is the ability to sort of amalgamate all of that knowledge into an analysis that you perform, both consciously and unconsciously.” Sommer and Zemanovich used connoisseurship to find small differences in brushstrokes, textures, lines and other details to determine the differences in small details between real and fake paintings. AI changing the art authentication landscape But the landscape of verifying paintings as real or fake is shifting now with artificial intelligence (AI) in the mix. Art authenticators are now able to use AI digital tools to find even more nuanced and finite differences between an artist’s authentic work and a copy. Recently, a Swiss AI firm called Art Recognition, used software to analyze disputed paintings attributed to Vincent van Gogh, and a Rembrandt painting called “the Polish Rider.” “AI is only as good as what you put into it,” said Anna Hudson, a professor of art history at York University. “So if you have the conservation information on Morrisseau’s paintings -- meaning the conservators, experts, collectors, connoisseurs of Morrisseau know there are certain colors, certain brushstrokes that characterize a painting -- it is true that one can get so familiar that you can you can really have your doubts, really feel strongly that a work is or isn’t legitimate.” “But AI is your second set of eyes that can search for those visual details, (and) analyze the materials. It is an ally in that regard,” she told CTV News in a Zoom interview on Sunday. But there’s concern that AI will also be used by copycat painters to create fakes that are tougher to tell apart from the real paintings. “I’m a pessimist because I’m immersed in a world where there are countless people who will take any opportunity to enrich themselves through fraud, and I know that if an opportunity presents itself to commit fraud in a more effective way, with a lower chance of being caught, and there’s money to be made from it, then they will do it,” said Sommer. “So, it’s not a question of ‘will this happen?’ I absolutely am sure that it will happen if it’s not already happening now,” he added. “This is why I see AI as much as it can be a helping tool, it could be the most dangerous thing that we’ve ever faced in the world of fraud.” The other nefarious way AI could be used is meshing it with robotics, Sommer said. This takes the shape of using automated machines, like robotic arms, to paint exact copies of artwork. “If you think of two systems within AI: you think of AI fraud detection systems, but you also think of AI-controlled robot painting systems and then you pit those two things against each other... you could be creating a kind of arms race between these two forces,” said Sommer. “The end result of that arms race can be a robot AI-controlled painting machine that can paint art that can’t be detected. Fake art that can’t be detected.” Using AI to copy paintings with artists’ permission Chloë Ryan, the CEO of Acrylic Robotics, has used AI and robotics to create art copies, but solely with the permission of the artist. Profits generated from copies are also sold directly to the original painter. “I call it the three C’s: consent, credit and fair compensation,” said Ryan in a Zoom interview with CTV News on Sunday. “We make sure that every artist that we work with is fairly credited. We also make sure that we have their consent, so we don’t train (AI) on any artist data ever without explicit written consent and a written agreement and compensation.” She believes this is an example of where AI’s ability to copy artwork can be used to benefit the artists, allowing them to make copies and generate revenue at a faster pace. “We’ve built the first technology in the world that can mimic an artist’s brush strokes using real paint on canvas, so an artist will typically paint digitally on a tablet interface, like an iPad or a drawing tablet. We capture their pressure, their speed, the pigment that they use and then you can imagine a large robotic arm dipping in paint and actually mimicking their brushstrokes on canvas.” Ryan said that, instead of selling one original for tens of thousands of dollars at a gallery, they can sell an entire licensed collection of quality replicas that allows the artist to make “a decent living,” while also spreading their artwork to a larger audience. Ryan believes that the benefits of introducing digital developments in the world of art will eventually benefit all of those involved. “That being said, it’s not lost on me that with great power comes great responsibility,” added Ryan. “As we continue to innovate in building this technology that gets better and better at mimicking an artist’s brush strokes, we have to stay on the cutting edge of fraud detection and making sure that our technology is not being used for nefarious reasons or with bad actors.”
