Killing Satoshi Axes Shooting on Location for AI

In partnership with

Welcome to VP Land! Doug Liman is shooting a narrative feature with Pete Davidson and Casey Affleck on a markerless capture stage in the UK. No real locations at all, every background AI-generated. The casting notice's fine print on consent and synthetic performances is where it gets interesting.

In today's edition:

Killing Satoshi builds every set with AI
AMC opts out before AI short ever screens
A-list talent stops fighting AI's inevitability
Google puts music generation inside Gemini

Doug Liman's Bitcoin Film Uses AI Stage

Doug Liman's Killing Satoshi, starring Pete Davidson and Casey Affleck, will shoot entirely on a markerless performative capture stage in the UK with no real locations, according to a UK casting notice reported by Variety.

All backgrounds and scenery will be AI-generated. The casting notice states producers reserve the right to "change, add to, take from, translate, reformat or reprocess" performances using generative AI, including adjustments to lip, facial, and body movements.
No AI-generated actors. Producer Ryan Kavanaugh told Variety: "We will not have any AI-generated actors that do not exist. AI is a tool we're using to make the filmmaking process more efficient while maintaining all department heads' jobs, all actor jobs." A reference to AI actors in the casting notice was included in error, according to a production source.
Union implications are significant. The production's approach sits at the center of SAG-AFTRA's ongoing negotiations with studios over synthetic performances, name/image/likeness protections, and consent frameworks.

Nick Schenk (Gran Torino) wrote the screenplay. The film follows the pursuit of Bitcoin's pseudonymous creator Satoshi Nakamoto. Production begins late February.

Worth watching: This is one of the first narrative features from a major Hollywood director to fully commit to AI-generated environments and AI-assisted performance adjustments from the start. The consent framework baked into the casting notice and the union response will set precedent for how productions handle synthetic performances going forward.

SPONSOR MESSAGE

6 AI Predictions That Will Redefine CX in 2026

2026 is the inflection point for customer experience.

AI agents are becoming infrastructure — not experiments — and the teams that win will be the ones that design for reliability, scale, and real-world complexity.

This guide breaks down six shifts reshaping CX, from agentic systems to AI operations, and what enterprise leaders need to change now to stay ahead.

Download the Playbook

Backlash kills AI short film run

An AI-animated short film won a festival prize that included a national theatrical run, but it never made it to the screen.

Screenvision Media, a third-party company that buys pre-show ad time at chains including some AMC locations, announced that the award winning short Thanksgiving Day would receive a national two-week theatrical run as its Frame Forward AI Animated Film Festival top prize. Before the film ever screened, online backlash erupted, and AMC proactively opted out.

The inaugural Frame Forward AI Animated Film Festival awarded its top prize to Igor Alferov's Thanksgiving Day, an AI-animated short made with Gemini 3.1 and Nano Banana Pro.

After online outrage, AMC opted out. "AMC was not involved in the creation of the content or the initiative and has informed Screenvision that AMC locations will not participate," the company said in a statement.

But they’ve played AI films before: AI content has played in theaters before, including at AMC. Runway's 2025 AI Film Festival screened in IMAX at 10 theaters.

Multiple A-list names across music and film are publicly accepting AI's inevitability -- a notable shift from the moral opposition that dominated the 2023 strikes era. Where the conversation two years ago centered on existential resistance, the tone from high-profile talent is moving toward pragmatic acceptance: AI is coming, so own your position before it arrives.

Ice-T told fans to accept AI-generated music videos as the new reality. Responding to a fan who argued AI shouldn't touch the music industry, the rapper and actor laid out a blunt economic case: "Fans want us to make and produce the music. Then shoot an expensive Video.. Then they get it for Free if they have an Apple subscription. Or Spotify pay us .007 cents a stream. The days of the expensive Videos are over. There isn't even MTV. Ai is the only sensible way to add visuals to a song. You can hate it all you want. It's the Future."

That same pragmatism showed up in a CNN/Variety interview, where Matthew McConaughey and Timothée Chalamet addressed AI in acting.

McConaughey predicted AI Oscar categories within five years. "Will we, in five years, have Best AI Film? Best AI Actor? Maybe. I think it could become another category. I'm not sure. It's going to be in front of us in ways we don't even see. It's gonna get so good we're not gonna know the difference."
He framed AI as a question of reality itself. "That's one of the big questions right now: the question of reality. It's more hazy than ever — in a very exciting way ... but also a scary way. Prep for it. Own your own lane, so you at least have agency when it starts to trespass."
Chalamet acknowledged the inevitability but stressed protection. "I'm fiercely protective of actors and artists in this industry. And equally, whatever tide is coming, it's coming." He called on Gen Z to lead AI integration while those in positions of power keep doors open for human talent.

The takeaway: The conversation has shifted from "should we?" to "how do we protect ourselves when it happens?" McConaughey's advice is concrete: own your own lane, maintain agency. Whether that's enough remains an open question.

New tools: rendering, editing, image gen, music

Four releases worth tracking:

Tavus Phoenix-4. A real-time human rendering model running at 40fps at 1080p. The hybrid Gaussian-diffusion architecture controls 10+ emotion states, active listening behaviors like nodding and concern, and full facial movement. It works alongside Tavus's Raven-1 (perception) and Sparrow-1 (timing) models. Available through the Tavus platform and APIs.
FireRed-Image-Edit-1.0. An open-source image editing model from the Xiaohongshu (RedNote) team. It claims state-of-the-art results on GEdit, REDEdit-Bench, and ImgEdit benchmarks among open-source models. The model is a ~28B parameter diffusion transformer, licensed under Apache 2.0, and available on Hugging Face and ModelScope.
ByteDance's autoregressive image model. ByteDance released a new open-source image generation model built on an autoregressive architecture, which the company says gives it stronger world understanding. The approach is similar to what powers Nano Banana and GPT-Image.
Google Lyria 3. Google's music generation model is built into Gemini and creates 30-second tracks from text or image/video prompts, complete with vocals, lyrics, and cover art generated by Nano Banana. All tracks carry SynthID watermarking. Available to 18+ users in all Gemini countries.

We have a full breakdown of what Lyria 3 can do.

How to Set Up OpenClaw

A guide on X (8,000+ likes) by @jordymaui distills 80 hours and $800 in Anthropic API tokens into a practical OpenClaw setup framework. The key lessons for anyone building an always-on AI assistant:

Use Claude Max, not the API. The pay-per-use API burns through credits fast. A Claude Max membership gives predictable costs and better throughput for agent workloads.
One agent with proper skills beats multiple agents. The author tried complex multi-agent setups before finding that a single well-configured agent outperforms a fleet of specialized ones.
Mac Mini as always-on hardware. Dedicated hardware keeps the agent running 24/7 without tying up a primary machine.
SOUL.md and USER.md are critical. These configuration files define the agent's personality, capabilities, and context. Getting them right is the difference between a useful assistant and an unfocused assistant.

The full guide covers setup steps, configuration patterns, and the mistakes to avoid. If you've been curious about running your own AI assistant, this is a solid starting point.

Brandon Lerry walks through a clay render to AI texturing workflow for creating hyper-real, cinematic 3D renders.

Stories, projects, and links that caught our attention from around the web:

🎬 John Gore Studios acquired a majority stake in Deep Fusion Films, a Cardiff-based AI production specialist founded in 2023 that builds ethically governed, insured AI tools for documentaries, factual entertainment, and scripted content.

🤝 OpenAI hired Charles Porch, Instagram's VP of global partnerships, as its first VP of global creative partnerships, a newly created role focused on building relationships with talent and creators across music, film, fashion, and sports.

🔧 Foundry (makers of Nuke) acquired Griptape, an enterprise AI orchestration platform that supports OpenAI, Google, and HuggingFace models, with plans to integrate it into Nuke and VFX/animation pipelines.

💾 iodyne expanded its Pro Data portable RAID line to 96TB and 192TB capacities, offering 80TB and 160TB usable under RAID-6 in the same backpack-ready form factor.

💰 Fei-Fei Li's World Labs raised $1 billion from AMD, Autodesk, NVIDIA, Fidelity, Emerson Collective, and Sea to build Marble, a product that generates persistent 3D worlds from images, video, or text.

👻 Meta patented AI technology that would keep deceased users' accounts active by replicating their behavior from past data, including posting, messaging, and video calling.

📺 Vū's One Mini LED wall now ships fully assembled and roll-out-ready, requiring one-person deployment with no build, starting at $99k.