China’s first fully AI-generated theatrical film opens to $181,000, Report Finds Midjourney Is Worst Offender for Reproducing Hollywood Copyrighted Characters, Galleri5 Launches AI Filmmaking Platform
China’s first fully AI-generated theatrical film has opened to a weak $181,000, offering an early measure of whether audiences will pay for synthetic filmmaking at scale. The result lands as Hollywood’s copyright fight with generative AI intensifies with a new report on offenders. Elsewhere, TIFFCOM is spotlighting artist-led AI and AI-produced series Space Nation and Kira premiered completed pilots to buyers at MIPCOM.
China’s first fully AI-generated theatrical film opens to $181,000
Sanxingdui: Future Past, billed as China’s first fully AI-generated movie to receive a theatrical release, has opened poorly at the Chinese box office. The 100-minute film earned $181,000 after opening on October 23 at 10,000 screens. Its producer, Bona Film Group, said the movie cost $1.5 million and took more than two years to make with a team of more than 100 people. Set in an algorithm-dominated future, Sanxingdui: Future Past used AI-generated characters, performers, and backgrounds.
GLIU Media & Entertainment is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
The result is an early commercial test for fully AI-generated feature filmmaking in a major theatrical market. Despite the low budget by mainstream standards, the film failed to convert into ticket sales. The “flop” contrasts with the unexpected performance of Niu Lai, a $200 animated film made by Xin Yumeng and his mother, Sun Lifang. The goofy-looking human-made film has reportedly earned $10 million.
Report Finds Midjourney Is Worst Offender for Reproducing Hollywood Copyrighted Characters
Midjourney’s AI model was the worst performer among 15 popular generative AI tools tested for reproducing copyrighted Hollywood characters and violating studio brand-safety standards, according to a report by copyright-testing startup VN.ai. VN.ai tested the models last month and found that Midjourney v8.1 generated images of copyrighted characters in harmful contexts once in every three text prompts, or 37% of the time.
The company defines brand-safety violations as cases in which characters are placed in contexts such as violence or drug use. One example cited in the report showed the Joker smoking cannabis while wearing a dress resembling Elsa’s costume from Frozen. It also found that every model tested generated at least one copyright-protected character, even when the character’s name was not included in the prompt. Google’s Nano Banana and ByteDance’s Seedance 2.5 were among the models that did so.
Seedance 2.5 improved significantly after receiving cease-and-desist letters from Hollywood studios. Other Chinese AI models continued to reproduce studio intellectual property. VN.ai said MiniMax H3 generated videos that violated studio brand-safety standards 55% of the time when an image was used in a prompt. Disney and NBCUniversal sued Midjourney last year, calling it a “bottomless pit of plagiarism.” The case remains ongoing, and Warner Bros. Discovery has also sued the company. Midjourney has argued that training its models on copyrighted images is permitted under fair use.
VN.ai founder Nelson Chu said models continue to use recognizable intellectual property because it is “good for business” and helps attract users. He predicted that behavior will improve as more companies establish standards for responsible use.
TIFFCOM 2026 to Spotlight Artist-Led AI at Tokyo Content Market
TIFFCOM 2026 will examine how artificial intelligence, machine learning and other emerging technologies can support storytellers rather than replace them when Asia’s leading content market returns for its 23rd edition from Oct. 28–30. The three-day event will be held at the Tokyo Metropolitan Industrial Trade Center Hamamatsucho-Kan, bringing together film, television and animation rights holders, buyers, producers and executives from around the world.
A keynote featuring AGBO partner Chris Brearton and chief scientific officer Dominic Hughes will focus on artist-led uses of AI and how technology can be integrated into the creative process. AGBO, the technology-enabled studio founded by Russo brothers directors Anthony and Joe Russo, has placed innovation at the center of its approach to next-generation storytelling. Brearton, who leads the studio’s business operations, and Hughes, whose background spans algorithms, machine learning and AI, will discuss how emerging tools can become creative partners for artists.
The wider TIFFCOM program will combine industry conferences with project and intellectual property markets, pitching sessions and networking opportunities. The event is designed to facilitate international co-productions and content business by bringing rights holders and potential buyers into the same marketplace.
Netflix Panel Says Human Touch Remains Essential to Filmmaking in AI Era
More than 300 Netflix titles use artificial intelligence in some form, according to a panel hosted by the streaming company at the Creative Asia forum during the Busan International Film Festival. Sung Q Lee, Netflix’s senior director of APAC production, said AI is used in previsualization, production, editorial, and visual effects. Lee said the technology can speed up production and help address practical challenges, but stressed that human involvement remains necessary. “AI gets the job done, but does it touch human emotion? The human touch is critical and necessary,” he said.
South Korean director Kang Yun-sung, whose credits include Run to the West and The Outlaws, said AI had reduced filmmaking costs and helped streamline production. Yun-Sung said he saved up to 40% of the budget on his upcoming film Gulf of Aden by using AI during development.
Weta FX Chief Technology Officer Kimball Thurston described the company’s approach as “assistive AI,” saying the technology should not change a filmmaker’s creative intent. Thurston said AI should not be treated as a single, uniform technology.
Netflix has previously said that roughly 300 of its titles have used AI at some point in their workflows.
AI-Produced Series ‘Space Nation’ and ‘Kira’ Premiered Pilot Episodes at MIPCOM
Two science-fiction series produced with generative AI, Space Nation and Kira, premiered their pilot episodes to buyers at MIPCOM in Cannes on Oct. 12 and 13. Both series were produced by Ex Machina Studios and are being sold worldwide by K5 International. The companies said the screenings would be the first time fully produced AI-generated series episodes, rather than concept reels or previews, were shown to buyers at a major international market.
Space Nation is an eight-episode series of 45-minute episodes set in the near future, when humanity’s survival depends on a new generation of elite pilots facing an alien threat. It was created by Ex Machina co-founder and CEO Marco Weber and Martin Weisz, who directs all eight episodes. Vanessa Coifman wrote the series. Utopia Studios and Globoplay are co-producers, and Globoplay acquired Brazilian rights, making it a Globoplay Original. Kira is a six-episode French sci-fi fantasy series of 35-minute episodes created and directed by Damien Steck and Julien Petit. It is set on the living planet Djena, where a young girl searches for her missing father while navigating rival factions.
The screenings come as K5 International represents a slate of AI-generated series and feature films, including Roger Avary’s Paradise Lost and Alex Proyas’ Heaven. K5 CEO Daniel Baur said both series were filmmaker-driven projects with completed pilots, strong scripts, and distinct worlds.
Galleri5 Launches AI Filmmaking Platform Targeting Global VFX and Animation Market
Galleri5, an AI-native technology company backed by Collective Artists Network, has launched enhanced capabilities for its AI Studio platform, positioning it as an alternative to traditional visual effects, animation, and film production workflows. The company introduced the technology at NXT in AI, its inaugural technology event in Hyderabad, attended by more than 150 filmmakers, technology leaders, VFX professionals, entrepreneurs and investors.
Galleri5 AI Studio combines more than 200 AI models with proprietary agentic workflows, character consistency systems, AI-powered visual effects and cinematic video generation tools. New features include Movie Agent, which takes filmmakers from story and visual development through storyboarding and shot generation; AI VFX, which transforms live-action footage and creates environments; AI Avatars for digital human creation; and production pipelines combining generative AI, 3D, animation and compositing.
Galleri5 has produced more than 10,000 commercial AI videos across advertising, television and cinema. The company is working with global AI technology partners to integrate advanced generative models into its workflows. Founder and CEO Rahul Regulapati said models are becoming available to everyone and that differentiation will come from workflows, tools and production systems.
Filmmakers Made AI Sci-Fi Series ‘The Arc’ With $38,000 in AI Credits
The creators of The Arc, a five-episode AI-generated science-fiction series, said the 83-minute production was made with about $38,000 in AI credits supplied by Artlist. The series was directed by Darin Stewart and executive produced by AI content creator Simone Ferretti. It was made with and greenlit by Artlist, a subscription platform that provides access to multiple AI image and video generators. The Arc is set in a near-future America and follows an incarcerated veteran who must fight through deadly historical simulations to buy a cure for his dying son. The series contains no human actors, sets, props, or costumes.
For each episode, they recorded a 10-minute verbal outline covering plot, dialogue, and camera movements, then used it as a prompt for Anthropic’s Claude to produce a working script, which they edited. They used ShotDeck and Figma to create character sheets and mood boards, and Google’s Nano Banana Pro and GPT Image 2 to generate images. Scenes were generated primarily with Seedance and Kling AI video models.
Each episode cost an average of $7,600 in AI credits and ranged from 10 to 22 minutes. Stewart said individual shots sometimes required eight to 10 generations. The team initially generated footage at lower resolution before increasing quality once it achieved the desired result.
The series had in-person screenings in London and Los Angeles.
