AI Vindicates Hidden Caravaggio: The Lute Player Emerges From Copy Status

29 September 2025

After decades of being dismissed as a derivative work, a painting long labeled “after Caravaggio” has now been attributed to the master himself through the revealing scan of machine intelligence. The artwork in question known as The Lute Player and housed at Badminton House was acquired in 2001 for roughly £71,000 by art historian Clovis Whitfield, who suspected it held deeper significance. Recent tests by the Swiss firm Art Recognition in collaboration with Liverpool University have produced an 85.7 percent probability that the painting is a genuine Caravaggio.

Previously, this version was overshadowed by a version in the Wildenstein collection, which for years had been accepted as authoritative. Institutional voices, including Keith Christiansen of the Metropolitan Museum, had challenged the plausibility of the Badminton version’s authenticity. Those stances are now a subject of renewed debate. The AI model rejected the Wildenstein version as non-authentic, suggesting mismatches in technical detail and compositional logic.

Support for the AI conclusion has come from unexpected quarters. Lute-maker David Van Edwards raised concerns about the Wildenstein instrument’s structural errors, while barrister William Audland framed his view in legal terms: that the cumulative evidence supports Badminton’s claim over the Wildenstein version. The finding, he wrote, challenges entrenched connoisseurship and calls for humility in the face of objective forensic tools.

The painting’s provenance adds a compelling layer. The Chigi collection—and earlier the Del Monte collection are historically tied to the composition, and early biographer Giovanni Baglione described a version so similar it matches details in the Badminton version: reflections in water, dew on flowers, and other minute visual cues. Whitfield himself cited that Baglione’s description aligns almost too precisely to be coincidence.

What happens now is less certain. If the attribution is broadly accepted, this painting would join the short roster of Caravaggio’s known works, transforming its value and place in art history. Whitfield has expressed a desire to see it placed in a public collection. Yet, full acceptance requires more than AI judgement. Critics still demand more conventional technical analysis infrared reflectography, pigment assays, brushwork study to corroborate what the algorithm suggests.

This moment sits at the intersection of art, science, and authority. The result is a reminder that as AI tools advance, they don’t just confirm what was once believed they force us to reconsider what we thought we knew. For centuries, attribution depended on expert eyes and subjective judgment. But now we live in a moment where pixels and pattern matching may wrest some of that decisional power.