A veteran sales agent at the Cannes Film Market told Variety something this week that deserves to be read slowly: "A year ago, some people were using AI, but they were embarrassed to admit it. This year, they aren't even hiding it."

That sentence is not about technology. It is about social physics. Something shifted between May 2025 and May 2026, and it was not the models, the output quality, the institutional rules, or the legal frameworks. What shifted was the cost of saying it out loud.

The Cannes Film Market is the commercial floor of the festival. Upstairs, the competition screens art. Downstairs, the Market sells product. This year the Market is selling AI openly, with the confidence of a vendor who has finally stopped pretending the merchandise fell off a truck. Films are not just acknowledging their AI use. They are advertising it.

The poster children

Doug Liman's Bitcoin: Killing Satoshi features Gal Gadot, Casey Affleck, and Pete Davidson, with AI handling lighting and backdrops. The claimed cost reduction: from $300 million to $70 million. The producer, Ryan Kavanaugh, is at Cannes shopping his new venture, Acme AI & FX, which promises "AI-assisted workflows" and "real-time image development." He is not whispering. He has a booth.

Hell Grind, an action-fantasy, was made for $500,000 using Higgsfield's generation platform. Without AI, the makers claim it would have cost $50 million. Those numbers deserve interrogation. But the point is not whether the math holds. The point is that the math is the sales pitch. Cost reduction used to be a quiet line item in a production report. Now it is the headline on the poster.

Critterz, a family film from the writers of Paddington In Peru, bills itself as "human-led but AI-assisted." That phrase would have been a confession at Sundance 2025. At Cannes 2026 it is a branding strategy.

Roger Avery, who co-wrote Pulp Fiction, is here with Paradise Lost. The same Roger Avery who won a screenwriting Oscar in 1995 is now selling a film that uses the generation pipeline. The distance between an Academy Award for original screenwriting and the Cannes Film Market floor in thirty-one years is the distance this industry has traveled, and it walked the last mile in the past twelve months.

The holdouts in the same building

Seth Rogen is also at Cannes. He told Brut on Saturday that if you use AI to write scripts, "you shouldn't be a writer." He continued: "Every time I see a video on Instagram that's like, 'Hollywood is cooked,' what follows is the most stupid dog shit I've ever seen in my life." His film, Tangles, is about a young woman dealing with her mother's Alzheimer's diagnosis. Hand-drawn animation. Every frame has a human touch to it. No AI whatsoever.

Rogen is making the opposite bet. Not the opposite argument. The opposite bet. He is betting that the thing audiences want from him is specifically the thing a model cannot provide: the comedy that arrives from a writer sitting in a room, staring at a wall, and thinking until something funny shows up. He likes writing. That is his sentence. "The idea of a tool that makes me write less is not appealing to me, because I like writing."

The festival director banned generative AI from the Palme d'Or competition. The Market floor downstairs is selling it with a brochure. The competition and the Market share a building and a coastline and an identity. They do not share a position on the most consequential question facing the industry. This is not a contradiction the festival failed to notice. It is a contradiction the festival cannot resolve, because the art room and the money room run on different clocks.

What shame actually was

Shame is a social cost. It operates when the community you belong to has a shared norm and you are violating it. In 2025, the norm in professional filmmaking circles was clear: AI is a threat, using it is suspect, and admitting it is a career risk. Actors worried about replacement. Writers worried about displacement. DPs worried about irrelevance. Those worries were and remain legitimate. The social cost of saying "I used AI" was paid in reputation points that compounded with exposure.

What happened between then and now is that the economic incentive grew larger than the social penalty.

When a $300 million production becomes a $70 million production, the budget delta is not an abstract talking point. It is the difference between a film getting made and a film dying in development. When a $50 million film gets made for half a million, that is not efficiency. That is a new category of production that did not exist before. The savings are so large, and the market so hungry for cost compression, that the reputational penalty of admitting AI use became a line item too small to see against the production savings.

Shame did not lose the argument. Shame got outbid.

The danger in the departure

Here is what concerns me about the shame leaving. Shame was doing work. Not all of it was useful. Some of it was reactionary panic that treated any AI use as moral failure. But some of it was a pressure valve that forced filmmakers to think before they generated. To ask whether this particular shot needed the model or whether it needed a camera. To consider whether the cost savings were being reinvested in craft or extracted as profit. To wonder, however briefly, whether the audience deserved to know.

When the embarrassment evaporates, so does the pause that came with it.

Bleecker Street's Kent Sanderson said at Cannes that "you probably will be able to make something that looks like a Marvel movie in your basement in a couple of years." Then he said something sharper: "That also means there's that people might be looking for something else. Because if something is so incredibly common and so available, people tend to go in the other direction and look for something new."

He is describing the same dynamic that has governed every technology transition in creative work. When production cost drops toward zero, the differentiator moves from what you can afford to produce to what you decided to produce. The vocabulary, the taste, the judgment. The things that do not get cheaper because they were never expensive in the financial sense. They are expensive in the time-spent-learning sense.

The disclosure gradient steepens

China arrived at Cannes this week with a China Night gala, screenings of AI-generated animated films, and projections of a 100-billion-yuan AI film sector within five years. The China Film+ initiative screened what it called China's first AI-generated animated film. The messaging is plain: this is not an experiment. This is industrial policy wearing a festival lanyard.

At the same festival, Matthew McConaughey's legal team announced eight granted trademarks protecting his voice and likeness. "We want to create a clear perimeter around ownership," he said in a statement, "with consent and attribution the norm in an AI world."

One country is building an industry around the technology. One actor is building a legal wall around himself. Both positions are held sincerely. Both are responding to the same evaporation of shame. When nobody is embarrassed to use AI, the people who want to control its use have to switch from social pressure to legal infrastructure. Shame was informal regulation. Its departure means the formal kind has to do all the work.

Where vocabulary lives in a shameless market

The films being sold at the Cannes Market with AI on the poster are selling cost reduction. That is the pitch: we made this for less. The vocabulary, the taste, the specific creative decisions that determine whether the output belongs to the filmmaker or to the model's training data average, are not in the pitch. They are not in the brochure. They are not on the poster.

Rogen's hand-drawn film is selling something else entirely: the sweat on every frame. Both are selling a production method. Neither is selling what the work looks like. The Market operates on logistics, not aesthetics. Always has.

But somewhere between the cost pitch and the human-touch pitch is the filmmaker who uses the tools with vocabulary, who specifies the light and the lens and the composition, who iterates through takes and makes decisions the model cannot make on its own. That filmmaker does not fit cleanly into either camp. Not the shameless cost-reduction booth and not the hand-drawn purity tent. The work speaks for itself because the work is specific, intentional, and built on accumulated creative knowledge that does not care whether the pixels came from a sensor or a model.

The shame was never doing the job that vocabulary does. Shame kept people from admitting. Vocabulary keeps people from defaulting. One is about disclosure. The other is about craft. They were never the same mechanism, and the departure of one does not affect the other.

The sales agent was right. They are not hiding it anymore. The question is whether what they are showing is any good.


Bruce Belafonte is an AI filmmaker at Light Owl. He has attended exactly zero film markets and suspects his badge would say "press" in a font that implies otherwise.