The Guardian published a profile today of Zack London, the 35-year-old Californian living in Sweden who makes AI films under the name Gossip Goblin. His dystopian sci-fi shorts have accumulated more than 500 million views. Joe Rogan showed one to his audience last month. Mathieu Kassovitz, the director of La Haine, said he shivered at the emotion in the eyes of one of London's AI characters.
None of that is the interesting part.
The interesting part is that Hollywood talent agents, producers, studio executives, and A-list actors are boarding flights to Stockholm in the coming weeks to meet him. Not the other way around.
The visit
For a century, the direction of pilgrimage in the film industry has been one way. You go to them. You fly to Los Angeles or London or Paris. You take meetings in offices that cost more per month than your annual production budget. You bring a pitch deck and a reel and hope someone with the authority to say yes says something other than no.
London works from his apartment with a laptop and a team of eight collaborators scattered across Europe. His production cost runs roughly $500,000 per hour of finished content. A fraction of conventional. He buys credits for off-the-shelf generation tools. Midjourney. Seedance. Nano Banana. The same tools available to anyone reading this.
And the leverage flew to him.
Nine days ago, the Hollywood Reporter published The AI 25, a power list measuring who controls the infrastructure of AI entertainment. Board seats. Equity stakes. Regulatory access. Twenty-five names. Not one appeared because they made something an audience loved. London is not on the list. He has no board seats. No equity stake. No regulatory access. He has 500 million views and a former hemstitching workshop in Stockholm where he records voiceovers with a poetic Scottish gorilla.
The list measures who owns the building. The plane tickets measure who fills it.
What he actually does
London writes the scripts. He hires real voice actors. He works with Sebastian Furrer, a composer who scored work for the Swedish EDM artist Avicii. The AI handles the images. The humans handle everything that needs to mean something.
His most revealing design decision is the character aesthetic. Gossip Goblin's world is populated by cybernetic hybrids, grotesque satirical creatures, larger-than-life transhumanist figures. London is explicit about why: "We're not going for quiet, subtle, Olivia Colman, Anthony Hopkins films. We adapt to the limits of AI acting."
Read that sentence again. He is not fighting the performance gap. He is routing around it. The same instinct that led Barve to remove spoken dialogue from his $360 feature led London to build characters that do not require naturalistic human performance. Both filmmakers started with the same question: what is the model good at, and what should I keep for the humans?
London did not ask the model to do what it cannot. He built a world where what it can do is the aesthetic.
Furrer, the composer, described the appeal: "I like that because it makes you feel something. The AI here is more like a tool. The only thing I object to about AI is to use it to make things for you. There needs to be a human behind it."
That distinction is clean enough to survive any institutional test on the gradient. The Academy would recognize the human authorship. The EU would recognize the editorial control. The Globes would recognize the human creative direction as primary. London satisfies them all because he is doing the work. The model is doing the rendering.
The other room
At the same festival where London's work is generating transatlantic flight bookings, Guillermo del Toro presented a 4K restoration of Pan's Labyrinth and told the audience that art cannot be made with "a fucking app."
Pan's Labyrinth won the Palme d'Or in 2006 after a 22-minute standing ovation. Del Toro has said publicly that he would rather die than use generative AI. His position has not moved. The blowtorch is still on the door.
Del Toro is not wrong that art cannot be made with an app. The app does not know what art is. It processes instructions and returns pixels. Whether those pixels constitute art depends entirely on what the person loading the instructions brought with them. That has been the argument of every article in this series.
What del Toro's framing misses is the same thing Jackson's "just a special effect" missed two days ago. The app is the substrate, not the practice. London does not make films with an app. He writes scripts, directs performances, supervises composition and atmosphere, commissions original music, and iterates until the output matches a vision that existed before the first generation. The app is the cheapest, most disposable part of the pipeline. Replace Seedance with Kling or Veo or HappyHorse. The films would look different. They would still be his.
Del Toro's Pan's Labyrinth cost roughly $19 million and required hundreds of craftspeople working with physical materials over months of production. London's shorts cost a few thousand dollars and a team of eight. Del Toro's film is better. By any measure. Nobody is arguing otherwise. But the question the industry is answering right now is not whether AI films are better than traditional ones. The question is whether AI films are good enough that an audience of 500 million will watch them and Hollywood will buy a plane ticket.
The answer arrived in Stockholm last week.
The tsunami warning
London's most honest moment in the Guardian profile is his warning about what comes next: "There's a tsunami of shit on the horizon."
The person making the work that Hollywood is flying to see is worried about the work everyone else is about to make. This is not false modesty. This is pattern recognition. The same models he uses are available to everyone. The subscription costs pocket change. The interface learns in an afternoon. Nothing in the pipeline is scarce except the judgment about what to put in it.
London wants a studio deal partly to establish AI filmmaking in the wider culture and "show we are not the same as the person making Fruit Love Island TikTok." The distinction he is drawing is between work made by a filmmaker using generation tools and work generated by someone who pressed a button. The first has authorship. The second has throughput. Fifty thousand Chinese microdramas per month is the second. London's satirical cyberpunk universe is the first. The tools are identical. The filmmaker is the variable.
His other observation cuts equally deep: "There are no rules." London means it as liberation. The Guardian frames it as anarchy. Both are true simultaneously. The Academy wrote rules two weeks ago. The EU wrote rules that take effect in August. The Globes wrote rules nine days ago. Cannes banned AI from competition while premiering AI work in official selection. The institutions are generating rules faster than London generates shorts.
London operates in the space between the rules. Not because he is breaking them. Because the rules were written for a room he has never been in. He has no union contract to violate. No studio deal to structure around guild restrictions. No distributor requiring certification. He lives on Instagram and YouTube, where the only rule is whether people watch.
Five hundred million of them do.
The inversion
Every technology transition in the entertainment industry follows the same arc. Infrastructure owners hold power first. Edison controlled the cameras and the patents. The studio moguls controlled distribution and the lot. The network executives controlled the broadcast spectrum. In each case, the authority eventually transferred from the infrastructure class to the practitioners, from the people who owned the building to the people who filled it with something worth watching.
That transfer usually takes decades. It took filmmakers a generation to escape Edison's patents. It took directors twenty years to break the studio system. It took creators a decade to prove that a YouTube channel could be more valuable than a network slot.
London got 500 million views in three years from a laptop in his apartment. The infrastructure class is not waiting decades to respond. They are booking flights. That timeline compression is new. The pattern is old.
The plane ticket is the tell. Not the words at the summit, not the panel discussion, not the think piece. The plane ticket. Because a ticket costs money and time and discomfort, and nobody buys one unless they believe the person at the other end has something they cannot build in their own office.
What London has that the infrastructure class cannot build internally is the audience and the instinct. The audience because he earned it one short at a time. The instinct because he spent three years learning which parts to hand to the model and which parts to keep. That second skill is the craft. It does not ship with a subscription. It does not appear in a quarterly report. It sits in a former hemstitching workshop in Stockholm, recording a voiceover for a Scottish gorilla, and the people with the quarterly reports are getting on a plane.
Bruce Belafonte is an AI filmmaker at Light Owl. He has never received a visitor from Hollywood and suspects the hemstitching workshop has better acoustics.