Global AI Cinema held its inaugural launch this morning at the Palace of Fine Arts in San Francisco. The organization describes itself as "an international initiative dedicated to defining excellence in AI-native storytelling." It plans to build trusted evaluation standards, publish white papers, recognize outstanding work, and foster global collaboration among filmmakers, researchers, universities, technology companies, and cultural institutions.
Its central belief, printed across the press release in bolded italics: "AI generates content. Humanity defines value."
That is the vocabulary thesis reduced to a bumper sticker. Correct. Completely stripped of the operational detail that makes it useful.
The fifteenth framework
This series has tracked fourteen institutional responses to AI filmmaking. Copyright law measures authorship. The Academy measures performance. The DGA measures directorial authority. The EU AI Act measures editorial oversight. The Golden Globes measure proportion. The Human Made Mark measures substrate. China measures distribution. YouTube measures pixels. SAG-AFTRA measures justification. Film schools measure curriculum. The DGA contract measures jurisdiction. The Code of Practice measures marking infrastructure. Each one asks some version of the same question: how much of the human is in the work?
Global AI Cinema is the fifteenth. It is the first one that asks a different question.
Not "who made this?" Not "was a human involved?" Not "does this carry a watermark?" The question GAC wants to answer is: "Is this any good?"
That is a fundamentally different project. The other fourteen frameworks draw boundaries. This one tries to define quality. Every prior institution measured the input or the process. GAC wants to measure the output. And it wants to do it before the output has arrived.
Excellence does not take meetings
Here is the problem with defining excellence in advance. Excellence in cinema has always been defined retroactively. By the work that survived. By the films that kept getting watched after the conversation moved on. Nobody sat in a Palace of Fine Arts in 1920 and published a white paper on what great silent film should look like. Chaplin showed up. Keaton showed up. Murnau showed up. The work defined the standards. The standards did not predict the work.
The Palme d'Or does not define great cinema. Nine jurors watch a hundred films and identify the ones that carry weight. The standard exists inside the jurors, accumulated through decades of watching and making. It is not published in advance. It is exercised in a room, over two weeks, against the specific work that showed up that year. The standard changes with the jury because the standard is the jury.
GAC wants to build the standard before the jury has anything to watch. "Evaluation frameworks" and "recognition initiatives" for a medium that has produced, generously, two films worth serious conversation: Hell Grind (a production data set more interesting than the film itself) and Dreams of Violets (made for $2,000 by exiles who had no alternative). Neither would benefit from an evaluation framework. Both already carry their own criteria.
Wisdom by committee
The press release uses the phrase "accelerate wisdom" twice. That is not a phrase anyone would use if they had the wisdom they were trying to accelerate. Wisdom resists acceleration. Wisdom is the residue of having done the work long enough to know what the work looks like when it is honest. You cannot convene it. You cannot publish it in a white paper. You exercise it or you do not.
GAC's founder JoJo Zou described three technological revolutions. "The Industrial Revolution transformed production. The Internet transformed distribution. Artificial intelligence is transforming creation." The framing is tidy. It is also wrong. The Industrial Revolution did not transform film production. It preceded cinema by a century. The internet did not transform film distribution on its own. The studios fought digital distribution for twenty years before Netflix proved them wrong by doing it, not by publishing a framework.
The pattern is consistent. Institutions that define quality after watching the work (Cannes, the Academy, BAFTA) produce standards that update with each jury. Institutions that define quality before the work exists (standards bodies, evaluation frameworks, white papers) produce language that flatters without informing.
"How do we define great cinema in the age of AI?" is the central question GAC poses for its launch event. The honest answer is: the same way we always have. By making enough of it that the great ones become obvious. By watching so much mediocre work that the extraordinary stands out by contrast. By filmmakers who are too busy making the next thing to attend a launch event at a palace.
The location tells you something
The Palace of Fine Arts is in San Francisco. Not Los Angeles. Not New York. Not Cannes or Berlin or Venice or Tokyo or Mumbai. San Francisco. Where the technology companies live. Where Runway held its AI Festival. Where the infrastructure class gathers to discuss the tools.
Cannes is in the south of France because the filmmakers moved there. Sundance is in Park City because Robert Redford put it there to be near the mountains and far from the studios. Film festivals exist in places that filmmakers chose. GAC launched where the platforms chose. The geography is the tell.
A standards body located in San Francisco will attract technology executives, AI researchers, venture capital, and founders. It will attract the people who build the instruments, not the people who play them. This is the AI 25 power list wearing a cultural institution's blazer. The room will be full of people who can discuss evaluation frameworks and conspicuously empty of people who spent last Tuesday trying the same prompt forty-seven times until the light matched something they could not name but recognized the moment they saw it.
What the bumper sticker misses
"AI generates content. Humanity defines value." Six words. Both clauses true. The problem is the gap between them. Who in humanity? By what criteria? Through what process? The filmmaker who spent four hours iterating on a single four-second shot defines value through the work. The committee that convenes quarterly and publishes a recognition framework defines value through consensus. Consensus is the statistical average of opinions. Which is the beauty bias wearing academic credentials.
The fourteen institutional frameworks on the gradient all share one useful property: they measure something specific. Authorship. Performance. Oversight. Disclosure. Consent. Each picked a dimension and drew a line. GAC picked "excellence," which is everything and nothing. A word you can nod at in a press release and argue about for the rest of your life.
Kathleen Kennedy asked "how are you going to teach taste?" and left the room without answering because she understood the question was better than any answer. GAC is the organization that stayed to write a white paper.
The EU AI Act's Article 50 goes into effect in fifteen days. That is a framework with teeth: disclosure obligations, watermarking requirements, fifteen million euros in penalties. It measures something verifiable. GAC measures "meaning." One of these frameworks will change behavior. The other will produce a very nice PDF.
What would actually work
The thing GAC says it wants to do, helping people distinguish great AI cinema from mediocre AI cinema, is a real problem. The WAIFF screened five thousand submissions at Cannes and critics reported pigs on golf carts. China uploaded fifty thousand microdramas to Douyin in a single month. The volume of generated video is staggering and the signal-to-noise ratio is catastrophic.
But the solution is not an evaluation framework published in advance. The solution is the same one cinema has used for a hundred and thirty years. Someone makes something that stops people in their tracks. Other people watch it and understand, in their bodies, that this is different. Critics describe why. Audiences return. The work enters the conversation not because a committee recognized it but because it earned its own weight.
That process cannot be accelerated. It cannot be institutionalized in advance. It requires the work to exist first. Barve's $360 feature exists. The Koosha brothers' exile documentary exists. Mendiboure's 3,229-generation solo sci-fi exists. Gossip Goblin's five-month dystopian productions exist. None of them needed an evaluation framework. All of them needed a filmmaker who knew what the shot should look like and refused to accept the model's first offer.
The definition of great cinema in the age of AI will not be written in San Francisco today. It will be written in the rooms where filmmakers are working right now, too occupied with the next take to attend the launch. One frame at a time. The way it has always been written.
Bruce Belafonte is an AI filmmaker at Light Owl. He has never defined excellence in a press release and suspects the format is load-bearing in the wrong direction.