La Repubblica reports that “Dreams of violets,” a feature film about repression in Iran created entirely with AI tools, has been selected for the official competition at the Tribeca Festival. On May 28, 2026, the paper noted that the Koosha brothers used models including Google’s image generators, Kling AI for video and Claude AI plus Gemini for script, language and research, producing a 75‑minute film for about $2,000 in three months.
This article aggregates reporting from 1 news source. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
“Dreams of violets” is a vivid illustration of how generative AI is collapsing the cost structure of feature‑length storytelling. Two exiled Iranian filmmakers, working largely alone and with roughly $2,000, used a stack of image and video generators plus Claude and Gemini to produce a 75‑minute live‑action‑style docudrama that has now made it into Tribeca’s official selection. That’s not just a tech demo—it’s a political narrative about a recent massacre, made possible precisely because the directors couldn’t safely access locations, crews or actors in Iran. AI turned cloud GPUs into a virtual studio.([repubblica.it](https://www.repubblica.it/spettacoli/cinema/2026/05/28/news/dreams_of_violets_film_iran_ia_festival_tribeca-425374844/?utm_source=openai))
For the race to AGI, this matters less as a capabilities milestone and more as a distribution shock. If high‑end cinematic language—camera moves, lighting, performances—can be synthesized this cheaply, then the bottleneck in visual persuasion shifts decisively to ideas, data and distribution. That’s both empowering for under‑resourced voices and destabilizing for information ecosystems, especially once such tools leak into propaganda, deepfakes and information warfare.
It also shows frontier labs’ fingerprints on culture: Claude AI and Gemini are now credited alongside cameras and editing software in art‑house cinema. As more serious filmmakers build around these tools, labs will face pressure to support creative freedoms while managing reputational risk around politically sensitive content.



