New technologies also face other challenges. Machine learning-based analysis tends to be difficult when an artist’s work exhibits stylistic variations or inconsistencies. Additionally, there has been little effort to develop AI authentication tactics for works in media other than paintings. Abstract artists present unique challenges, especially when there is no expression of gesture or sign in their work. Still, several recent studies have reported success in using AI to determine the authenticity of Jackson Pollock’s works.
Based on current technology, AI can best be applied in three areas: When performing stylistic analysis, AI can successfully recognize patterns and nuances in an artist’s method, such as brush strokes, color palettes, and repetitive compositional tactics. In one recent study, AI studied the portrait poses of hundreds of works of art and suggested significant differences in head position (pitch, yaw, roll) across the sample set, allowing it to accurately determine whether a work was by a Japanese ukiyo-e or a modern neo-primitivist artist.
AI can also be successfully implemented in material analysis, comparing characteristics such as pigment type, binder, and canvas texture to ensure that the work is compatible with both the artist’s period of practice and specific characteristics. Finally, experts agree that AI can provide much-needed support in provenance tracking. With its ability to quickly and efficiently search and consider large datasets, AI has the potential to save time when reviewing historical records, auction catalogs, and other archival records.
In the area of provenance and ownership, non-fungible tokens (NFTs) have been touted as a potential deterrent to counterfeiters, providing a record of creation and ownership. NFTs tied to cryptocurrencies can protect against unauthorized copying and ownership issues that have plagued digital media. When an NFT is sold, the transaction is documented on the blockchain, an immutable archive of cryptocurrency transactions. Collectors of contemporary art may be most familiar with the rise of NFTs circa 2021. In March of that year, digital artist Beeple’s Everydays: The First 5000 Days sold for more than $69 million at Christie’s.
Several recent cases demonstrate how ownership records on blockchain can disrupt traditional ownership systems and provide a valuable resource for provenance investigations and disputes over authenticity. In 2021, digital asset bank Sygnum and non-traditional investment firm Artemundi partnered to create an Art Security Token (AST) for Picasso’s oil on canvas painting “Fillet aux Belles” (1964). Tokenization allows multiple patrons to jointly own a work. By the end of the subscription period, 50 investors had backed the painting, creating the first example of digitally authenticated co-ownership of a major work of art.
That same year, Damien Hirst sold 10,000 works, collectively known as Currency. Buyers had one year to decide whether they wanted to own the NFT or any of the associated unique works on paper. Unclaimed paper works are said to have been incinerated. According to the artist, the project challenges the value system within the market and, in turn, raises questions about what a real work of art is or can be.
Digital and AI-powered authentication practices are here to stay. Major institutions such as the Van Gogh Museum and the Louvre have recently invested in AI to identify historically challenging works in their collections. Private organizations and public scholars alike continue to document the capabilities of this technology. Nevertheless, as Hephaestus CEO Denis Moiseev asserts, AI is “not a silver bullet.” As digital humanities scholars Peter Bell and Fabian Offert have argued, relying solely on AI authentication will create a new formalism of the type that past art historians, such as 19th-century Viennese expert Alois Riegl, have rejected. Best practices in this field require a balance between formal analysis (a skill set in which AI excels) and contextual analysis (a learned and practiced skill that currently only human experts can perform).
