Generative AI Headlines
AI-Powered Digital Twins of Deceased Loved Ones, AI twins serving as mentors, tutors, or interactive textbooks, New Oscar Rules for AI, Arcana Labs AI Film Production Takes on Hollywood
In Europe, the Paris AI Policy Action summit in February failed to deliver significant regulation. In the United States, the new administration’s March directive from the National Institute of Standards and Technology removed references to “AI safety” and “AI fairness.” At the same time, Meta and other platforms have dismantled fact-checking operations, fueling a wave of AI-generated video content that is overwhelming the internet, with some pieces drawing hundreds of millions of views. This regulatory vacuum is raising alarms about media integrity and intellectual property. Concerns have prompted a flurry of responses to the forthcoming U.S. AI Action Plan, with Hollywood pushing for stronger protections against AI-driven copyright abuse. The debate is intensifying as the U.S. plan nears release and the European Union prepares to implement the AI Act.
AI-Powered Digital Twins of Deceased Loved Ones
The past few years have seen rapid advances in generative AI, particularly with large language models like GPT-4, Llama 2, and Palm 2. There has been accelerated growth in the advancement of AI-generated images, video, and audio, alongside a new generation of multimodal models that combine multiple media types into a single system. These developments have led to generative agents- AI systems that can mimic human behaviors such as memory and planning. Though still in the early stages, these agents are expected to grow more sophisticated and widely used as model capabilities improve, computing costs decrease, and technical expertise becomes more accessible.
Several companies provide AI-powered digital twins of deceased loved ones, with services from Character.ai, StoryFile, DeepBrain AI’s Re;memory, HereAfter, Project December, and MIT’s Augmented Eternity. These platforms use personal data, recordings, and digital footprints to create interactive avatars that mimic the voice, appearance, and personality of the departed. But researchers at Google DeepMind and the University of Colorado are pushing the concept further with “generative ghosts.” Unlike digital twins that simply recreate past behaviors, generative ghosts can hold new conversations, react to current events, and even simulate decision-making-essentially evolving beyond static memories. The technology is designed to generate responses and behaviors that feel authentically human, raising the possibility that these AI agents could continue to interact, work, or offer support long after someone has died. As these generative agents grow more advanced, experts predict it may soon be common for people to create custom AI versions of themselves to connect with loved ones and the wider world after death, blurring the boundaries between memory, identity, and digital immortality.
The technology is already finding real-world applications. Executives and thought leaders are building digital vaults filled with years of speeches, articles, social posts, and interviews, creating interactive, searchable knowledge bases that work for them and engage their audiences.
Twin Protocol specializes in these “vaults,” curating data from books, lectures, videos, and voice memos to train highly personalized AI twins. Using a proprietary persona AI layer, these twins capture an individual’s tone, style, and intent, resulting in an interactive version that responds with their knowledge and voice, not just a generic chatbot. As the market grows, Twin Protocol sees potential in education and research, where curated AI twins could serve as mentors, tutors, or interactive textbooks, bringing real voices into classrooms or letting students interact with digital versions of historical figures.
New Oscar Rules for AI
The Academy released new rules on AI usage in films submitted to the Oscars. Using generative AI won’t automatically boost or tank a film’s chances. What matters most is how much human creativity drives the project. The Academy says AI tools alone don’t decide nominations; it’s the human touch behind the scenes that counts. The Academy’s update reflects how AI is becoming a regular part of filmmaking, from voice tweaks to visual effects, without requiring filmmakers to disclose its use. Human creativity remains the deciding factor in what makes a film Oscar-worthy.
Arcana Labs AI Film Production Takes on Hollywood
As generative AI startups pour into Los Angeles’ entertainment scene, venture capital is quickly following. Arcana Labs, co-founded by Millennium Media president Jonathan Yunger and producer Hank Hoffman, emerged from stealth in April with $5.5 million led by SEMCAP AI. The funding will help launch its digital studio, Arcana AI, aimed at bringing AI-driven tools to film production. Arcana Labs got its start after Yunger and Hoffman experimented with Midjourney’s AI image platform and became hooked. The company connected with SEMCAP through its production arm, Primary Source, after helping troubleshoot an AI issue on a George Washington documentary series. Arcana’s platform now boasts over 100,000 users, and several major studios and streamers are piloting its tools, though details remain confidential. Visual effects teams and production designers are incorporating Arcana’s AI into their workflows, and the company recently signed a contract with SAG-AFTRA to guide responsible AI use in the industry.
Natasha Lyonne Makes Feature Directorial Debut with AI Sci-Fi Film
Jaron Lanier, the pioneering technologist and frequent AI skeptic, is joining forces with Natasha Lyonne and Brit Marling for a new feature film set in the world of immersive video games and powered by artificial intelligence. Titled “Uncanny Valley,” the project is backed by Asteria, the AI-driven studio Lyonne co-founded with filmmaker Bryn Mooser. Lyonne will make her feature directorial debut with a script she co-wrote with Marling; both will also star. The film follows a teenage girl who loses her grip on reality after becoming obsessed with a wildly popular augmented reality game in a parallel present. “Uncanny Valley” will blend live-action and video game-inspired visuals, with Lanier, Lyonne, and Marling collaborating on the game elements. The production will use Moonvalley’s “Marey,” a generative AI model built only on copyright-cleared data, distinguishing it from other AI tools in the industry. Asteria describes the film as an experiment in merging classic storytelling with cutting-edge AI, aiming to push the boundaries of what’s possible in cinema. While its release platform is still undecided, the project has already sparked industry buzz, with Lyonne and Marling positioning “Uncanny Valley” as a bold exploration of technology, creativity, and the human experience.
SAG-AFTRA’s New Commercials Deal Puts AI in Check
SAG-AFTRA’s National Board has approved its 2025 commercials contracts, sending the deal to members for ratification by May 21. The agreement is valued at $218.4 million in increased earnings and benefit contributions over three years. If ratified, the contracts would prohibit the use of performers’ data and likeness to train AI without consent, boost one-year streaming use fees by up to 33.33%, and raise paid YouTube.com rates by 12% over current digital rates. Performer compensation would rise by 5% in the first year, 4% in the second, and 3% in the third. SAG-AFTRA leaders called the deal a major step for wage growth and AI protections as the industry adapts to new technology and digital platforms.
Film & TV Marketing Gets a Boost from AI Eye-Tracking
AI-powered tools are increasingly being used by agencies to create official trailers, promos, and marketing materials for studios, streamers, and broadcasters. Once Upon a Time, an agency known for its award-winning work on films like Venom and series such as The Crown, is a leading example of how AI is influencing entertainment marketing. This increased use of AI-driven content follows a recent Deadline report exposing fake movie trailers generated by AI on YouTube, which led the platform to cut ad revenue for two channels monetizing the unauthorized videos.
The agency uses Dragonfly AI, a platform originally developed by UK academics that employs a patented algorithm to predict what elements of visual content will attract human attention. In entertainment marketing, these insights are used to fine-tune materials like trailers, posters, and key art for maximum impact across platforms. Notably, the practical application of Dragonfly AI typically happens after the main creative design is complete, offering data-driven refinements without replacing the creative process itself. This allows marketers to optimize visuals for different contexts while preserving the core artistic vision.
Dragonfly AI is applied post-design to optimize key movie and series visuals for different platforms. It adapts trailers and posters for theatrical release, on-demand, streaming, or broadcast, ensuring each asset performs across diverse channels. By tailoring materials, content creators can extend the lifespan and commercial reach of their IP.
Organizations need to prepare for a shift in their creative and design workflows as generative AI becomes a standard tool. Corporate materials, marketing pitches, branding assets, and other content can now be produced and distributed at unprecedented speed, but that’s only part of the story. Understanding the potential and implications of this technology is crucial. While some creators and designers remain wary of AI’s impact on their work and careers, others are embracing new creative opportunities and possibilities it brings.
Ginger Liu is the founder of Hollywood’s Ginger Media & Entertainment, a researcher in artificial intelligence and visual arts media, and an entrepreneur, author, writer, artist, photographer, and filmmaker. Listen to the Podcast — The Digital Afterlife of Grief.
Et tu, Jaron? We shall see.
I'm not a Luddite, but we've made gorgeous films for a long, long time without AI eye-tracking or fake vocals. We're sanding away humanity every single day in pursuit of VC money. One day, we'll look up and wonder why we (May 2025 humans) didn't stop it.