Few names in Renaissance art carry the weight of Raphael. Celebrated in his own lifetime and canonized soon after, Raffaello Sanzio occupies a rare position as both an artist of supreme refinement and a figure whose works were intensely desired, copied, adapted, and extended by others. That combination of fame, stylistic clarity, and collaborative production has made attribution in Raphael studies unusually complex, particularly for devotional paintings that survive outside firmly documented papal or court commissions.

The painting now known as the Flaget Madonna belongs squarely within that contested terrain. Discovered in 1995 in the English countryside, the work depicts the Virgin Mary cradling the Christ Child, observed tenderly by Saint Elizabeth and the young John the Baptist. An oak tree rises in the background, its branches hosting a goldfinch, a symbol long associated with the Passion and Crucifixion. The composition is immediately legible within Raphael’s visual world, both iconographically and emotionally. The question has never been whether the painting aspires to Raphael’s language, but whether Raphael himself had a hand in its execution, and if so, where that hand begins and ends.

The painting’s modern provenance begins with its association with the Sisters of Charity of Nazareth, a convent in Kentucky. According to early investors and individuals familiar with the work’s history, it was donated to the order by Bishop Joseph Benedict Flaget, one of its founders, from whom the painting takes its name. Beyond that point, documentation thins, leaving a gap that is not unusual for devotional paintings that circulated privately or within religious communities for centuries. It is precisely this combination of stylistic proximity and documentary silence that has fueled decades of debate.

Raphael’s career trajectory provides important context. His early works in Umbria and Florence were executed in oil or egg tempera, before he adopted oil painting fully as it became standard practice in Italy. After his move to Rome, his output expanded dramatically. Large fresco cycles for the Vatican required extensive assistance, and Raphael’s workshop grew into a highly organized operation, staffed by talented pupils such as Giulio Romano and Francesco Penni. Within this system, authorship was often distributed. Raphael might design a composition, paint key passages himself, and leave secondary areas to trusted collaborators. In other cases, workshop members produced works in his manner with minimal intervention.

This reality has long complicated attempts to draw firm boundaries around Raphael’s hand. When the Flaget Madonna was shown to art historians over the years, reactions clustered around a familiar set of possibilities. Some Renaissance scholars such as Patricia Trutty-Coohill, Marcia Hall or Larry Silver argued for autograph authorship, citing the tenderness of the faces and the psychological coherence of the grouping. Others proposed a collaborative work, suggesting that while the composition reflects Raphael’s influence, the execution lacks consistency. Still others entertained the idea of partial authorship, a hypothesis that is often difficult to substantiate with traditional tools alone. Scientific and forensic analysis provided an important first layer of clarification. Art Analysis & Research, the London-based forensic authentication firm, conducted technical examination of the painting, determining its approximate date and geographic origin and identifying the presence of orpiment —a yellow pigment used by only a handful of Renaissance artists dueto its high toxicity. One of those artists was Raphael. The forensic investigation ruled out other known artists of the period, including those often proposed as alternatives, such as Ceraiolo. The findings narrowed the field significantly, but they did not, on their own, resolve the question of authorship within Raphael’s orbit.

To address that remaining uncertainty, Art Recognition conducted an AI-based analysis focused on isolating authorship at the level of specific pictorial passages rather than issuing a single, global verdict. The system was trained on a comprehensive dataset of confirmed Raphael works, alongside a robust negative training set that included paintings by Raphael’s closest collaborators and contemporaries, including Giulio Romano, Gianfrancesco Penni, Perino del Vaga, Parmigianino, Sebastiano del Piombo, Polidoro da Caravaggio, and Giovanni Antonio Bazzi. Modern imitations and AI-generated images in the style of Raphael were also included to strengthen discrimination. The complete Raphael training dataset is publicly available here, allowing the methodology to be examined and tested independently. 

The results were striking in their specificity. The faces of the Virgin Mary and the Christ Child were classified as autograph Raphael with a probability of 97 percent, well above Art Recognition’s threshold for convincing authenticity. By contrast, the remainder of the painting was classified as not by Raphael. Rather than undermining the painting’s significance, this differentiated result offers a plausible explanation for why the Flaget Madonna has resisted easy classification for so long. Seen in this light, the painting aligns closely with practices in which more than one hand contributed to its execution. Raphael’s direct involvement appears concentrated in the most expressive and psychologically charged elements of the composition, where his hand is most often recognized by connoisseurs. Other areas, including secondary figures and background passages, display characteristics consistent with associate execution. The AI analysis does not contradict earlier scholarly hesitation. It explains it.

What makes this case particularly instructive is that it avoids forcing a false binary. The question is not whether the painting is “by Raphael” or “not by Raphael,” but how authorship was distributed within a collaborative system. Traditional connoisseurship has long acknowledged this possibility, but it has lacked reliable tools to map it with precision. In this instance, computational analysis supplies that missing granularity, distinguishing autograph passages from studio work in a way that aligns with historical practice.

The Flaget Madonna has attracted broader public attention as well, including coverage in The Wall Street Journal. That attention reflects a growing interest in cases that do not simply confirm or reject attribution, but deepen understanding of how great Renaissance workshops actually functioned. Raphael’s legacy was not built on solitary genius alone, but on a carefully managed network of artistic production. Recognizing that complexity does not diminish his achievement. It situates it more accurately.

This case demonstrates how AI, when trained responsibly and interpreted conservatively, can extend rather than flatten art historical judgment. It does not replace expertise. It sharpens it. By revealing where Raphael’s hand is most convincingly present, and where it likely recedes, the analysis invites scholars to revisit long-standing assumptions with new clarity. In doing so, it offers a more nuanced answer to a question that has never truly been simple.

< Back