A morning briefing mentioned a "random film of the day" — "How to Steal a Million" (1966) by William Wyler: Audrey Hepburn, Peter O'Toole, a Parisian museum heist over a "Cellini Venus" statuette. Usually such items roll by in the background: a sweet film from half a century ago, and that's it. What hooked me was the premise: to shield a family heirloom on loan to a museum from scientific analysis suddenly required by an insurance policy, the heroine is forced to steal it from the museum.
This isn't a comedy plot — it's a protocol for how the real art market operates. Forgeries pass not around experts but directly through their hands, and they're caught not by detectives but by chemists and dendrochronologists. All that remained was to check how closely the film matched reality — and it turned out this "silly comedy" has a precise real-life prototype, an entire school of followers, and a fairly grim sequel. Along with a question that's become commonplace in the era of generative models: what does "original" even mean?
The production history of the film is itself a detective story. Shooting began August 16, 1965 in Paris, and the city held up permissions for a long time: the plot politely hinted that forgeries might hang in the Louvre. Interior scenes were shot at Studios de Boulogne, location work in Paris, and the "Old Masters" from the father's collection were made by props departments after Gauguin, Cézanne, Degas, and Renoir — they "looked authentic," which the creators immediately boasted about to the press.
And they boasted twice, lying once. First a "technical advisor" told the British press that the prop fakes were indistinguishable from originals and would be destroyed by order of the French Ministry of Culture. The story flew through newspapers until the paintings were shown in Paris — where art critic Pierre Schneider delivered his verdict: "worse than mediocre." A lesson in honesty that the art market didn't absorb right away: a forgery is convincing only until it's placed next to an original.
A separate comedy is that the film's technical consultant was Joe Chapman — a former FBI agent. Wyler assembled a genuine forensic squad around this comedy. Audrey Hepburn took driving lessons to race a sports car through Paris traffic herself, her fee was a million dollars, and her wardrobe by Hubert de Givenchy cost 30 thousand, with color tests for each outfit. The budget by the end of shooting had grown to 6–7 million, George Scott was fired on day one (showed up at five p.m. — and Wyler, citing the fact that "the French industry starts the day at noon," instantly replaced him with Eli Wallach).
The box office, absurdly enough, fell into the same trap as the plot: by the studio's calculations the film should have brought in 12 million in rentals, collected 10.45 — and was listed as a loss. But critics later rehabilitated everything: the film has a rare 100% on Rotten Tomatoes, and Bosley Crowther in The New York Times wrote that the plot was "impossible," but it was "delightful nonsense — especially the central museum heist number." The "Papa! — Nicole!" exchange from the film was stolen thirty years later by Renault Clio advertisers, and the alarm scene migrates through cinemas on three continents — from Bollywood's Loafer to a 2019 Argentine comedy.
Screenwriters Harry Kurnitz and George Bradshaw probably weren't peeking at crime chronicles, but ten years before the premiere this plot played out in reality almost verbatim. Dutch artist Han van Meegeren, whom critics considered "a pale copy of his teachers," nursed a grudge against the experts and spent six years developing technique: he bought authentic 17th-century canvases, painted "lost Vermeers" over them, and mixed paints with phenol-formaldehyde resin — after hot drying, the bakelite "aged" the paint layer three hundred years.
In 1937 through a lawyer he brought Abraham Bredius — in the art world they called him simply "Papa" — the leading Vermeer expert, "Supper at Emmaus." Bredius wrote an article in The Burlington Magazine, "A New Vermeer," and declared the painting "a great masterpiece by Johannes Vermeer of Delft." Chemical tests, analysis of white lead, X-rays, and microspectroscopy "confirmed" authenticity — the Rembrandt Society bought the canvas for 520 thousand guilders (around four million in today's money) and transferred it to the Boijmans Van Beuningen Museum, where at Queen Wilhelmina's anniversary exhibition the fake became the "spiritual center" of the exposition.
During the occupation Hermann Göring bought "Christ and the Adulteress" from van Meegeren — giving 137 confiscated paintings and 1.65 million guilders for it. After the war the artist was arrested for selling cultural treasures to the Nazis with the threat of the death penalty, and then came the scene that the 1966 comedy retold almost word for word: "The painting in Göring's hands is not, as you suppose, a Vermeer of Delft, but a van Meegeren! I painted it!" It took six months to prove — the forger under supervision of reporters and court witnesses painted another painting "in the style of Vermeer," after which the collaboration charge collapsed: what he'd sold wasn't national heritage but a forgery. The court in November 1947 gave him a year for forgery and fraud; two months after the verdict van Meegeren died of a second heart attack.
This ending is impossible to make up: an October 1947 poll ranked him as the second most popular person in the country — right after the prime minister and just ahead of Prince Bernhard. The public saw in him a trickster who had fooled both snobby experts and the Reichsmarschall; by contemporaries' accounts, Göring, learning about the forgery, "looked as if he'd discovered for the first time that evil existed in the world." Van Meegeren himself summed it up more accurately than any film critic: "My triumph as a forger became my defeat as an artist."
The comedy is accurate in details that at first seem like hyperbole. The enemy of the forgery in the film isn't police or museum security but the fine print of an insurance policy: the heroine's father signed a standard document and didn't notice the clause about scientific analysis. That's how it was in life: the leading role in exposures is played not by detectives but by underwriters, lawyers, and tax authorities. And then the main market mechanism kicks in — reputation instead of verification.
This is explained not by theory but by the largest art fraud in American history. The Knoedler Gallery — the oldest in New York, with an impeccable name — sold "newly discovered" works by Rothko, Motherwell, Pollock, and de Kooning for fifteen years until the total reached 80 million dollars. There was one source: "international art dealer" Glafira Rosales, whom no one had the courage to ask for an address — private investigators found her in a week at a Long Island cottage with newspapers on the lawn. Pei-Shen Qian painted everything — an artist from Queens, in a garage; he got maybe 65 thousand when the turnover was 80 million, but by his account he never intended to fool anyone: he was selling "imitations," and for some reason people took them seriously.
The truth surfaced not through connoisseurship but through chemistry. Historian Jack Flam noticed that signatures on seven "Motherwells" were traced "as if by template," and the inscription "Spanish elegy" on the reverse the artist never used; forensic chemist James Martin found under a stereomicroscope traces of an orbital sander in the lower layers, and Raman spectroscopy — a pigment that appeared on the market only more than ten years after the paintings' "date of creation." One of the fake Pollocks, incidentally, came with a signature missing the letter "c." No magic: the arsenal that in the film an insurance expert was supposed to deploy on the "Venus" worked in reality exactly the same way — silently, boringly, and irrevocably.
The psychology of this Knoedler blunder was formulated by Flam himself: the catalogue raisonné is "the Bible of the market," and the Bible isn't verified. Bredius in 1937 dreamed of finding a lost Vermeer — and found exactly what he dreamed of; a gallery with a century-old reputation "couldn't" sell garage forgeries — and didn't verify. A fake doesn't need to deceive vision: it's enough to hit someone else's desire.
The tastiest thing in the real chronology is that the Louvre, which the film so delicately mocked, itself became victim of the same scheme. In 2010 Parisian dealer Marc Weiss and investment fund Fairlight bought an "early Frans Hals" for three million euros — while the Louvre was simultaneously preparing a purchase for five and had already declared the work "national heritage" with an export ban. It ultimately sold for 11.75 million dollars through Sotheby's; in 2016, amid a French criminal case against supplier Giuliano Ruffini, the auction ran the painting through forensics — 20th-century pigments sat deep under paint layers — and returned the full amount to the buyer.
The denouement stretches into a sequel: a 2019 London court forced Fairlight investor to compensate Sotheby's 5.375 million, the dealer settled for 4.2, and supplier Ruffini received a French arrest warrant. But the liveliest detail is the conflicting matrix of expertise: shortly after the test the auction house bought Orion Analytical laboratory and made its founder staff scientific director, for which in court it received the label "partisan expert." Judge Robin Knowles made a point worth its weight: the verdict "does not determine whether the work is by Frans Hals." Science confirms anachronism, but the market still hasn't closed the question "is this original or not."
The Basquiat case from the same years is a textbook example of how cheaply you can pass through institutions. In 2022 the Orlando Museum exhibited 25 "newly found" 1982 Basquiat works "discovered" in the late TV screenwriter's storage unit; the screenwriter himself had told investigators during his lifetime that he never owned a Basquiat. One of the drawings was executed on cardboard with the inscription "Align top of FedEx Shipping Label here" — that font version appeared in 1994, six years after the artist's death. This isn't subtle forensics, it's a stamp with a date.
The FBI raided the museum on a Friday, with visitors present, and hauled out all 25 works; in April 2023 ex-auctioneer Michael Barzman admitted that his partner drew the pictures — under a plea agreement, "maximum 30 minutes per image, sometimes five," after which they sold them on eBay. Worth separate mention is the prediction of the lawyer who bought a share in six works: "forgers make one big impressive fake, no one will go to the supermarket for cardboard for 25 paintings." Reality turned out exactly opposite: a cheap run "as street art" met the least resistance from experts — overly clever intuition was again looking for a serious enemy.
If van Meegeren is the plot's prototype, then Eric Hebborn is its dark sequel. A London draftsman with a Rome Prize, favorite of Sir Anthony Blunt (who noticed that a pair of drawings "resembled Poussin" — that's how it all started), Hebborn made forgery not a scam but a craft: old paper from antique shops, homemade inks, aged seals. National Gallery director Sir John Pope-Hennessy acknowledged his drawings as "authentic and stylistically brilliant," sold through Christie's and Sotheby's, and Hebborn himself claimed that over his lifetime he put thousands of fakes into circulation totaling over 30 million dollars.
He wasn't exposed by experts but by clerical oversight: in 1978 curator of the Washington National Gallery Konrad Oberhuber noticed that two drawings purchased at different times from dealer Colnaghi were executed on the same paper. Even after this the trading house was afraid to name the name aloud for eighteen months — fear of a libel suit proved stronger than professional ethics; "no one speaks of him... the trouble is he's too good," relayed an acquaintance of dealers. Hebborn responded with a textbook manual: the posthumous "The Art Forger's Handbook" — open documentation of the craft, from boiling paper to forging provenance.
He ended like a noir character: in January 1996 Hebborn was found on a Roman street with a skull fractured "by a blunt instrument" — days after the Italian edition of his handbook came out. The murder is unsolved. And the market reacted in a way that makes comments superfluous: in 2014 his 236 drawings went at auction for fifty thousand pounds, a sanguine "after Michelangelo" fetched a price eighteen times the estimate. The forger's forgeries became collectible items: the criminal's signature finally melted into authorship.
The comedy's plot closes with a perfect scene: "thief" Simon turns out to be a detective-consultant hired by galleries to catch forgeries — and the entire edifice of deception built in the film hangs not on crime but on who dares certify a thing with a signature. The final punch — Bonnet, disappointed that the insurance payout won't happen because the "Venus" was never authenticated, and a new "Van Gogh" behind the shoulders of a "cousin" from the next generation. No one was punished: demand for a beautiful legend is immortal.
Against this backdrop "How to Steal a Million" plays more accurately than any art noir: the only honest moment in the economics of art is when someone in gloves drives a statuette to a spectrometer. Generative models have only raised the stakes of this old game: when style is replicated for free, what remains scarce is not the image but attribution — and court battles over "authorship of machine labor" are already playing out in the same genre as the Ruffini case, only the proportions are different. The analogy here is not a verdict but an open question, and it's worth keeping in mind: every time the market encounters a "lost masterpiece," the most valuable thing in it isn't canvas or pigments but someone else's desire to believe.