The Big Story

A filmmaker just released the first complete feature-length movie made entirely by AI, and every single one of its 473,600 assets is now public for anyone to download and study. The invitation to watch came from a New York screening of this project, which stands out because it secured legal licenses for celebrity likenesses before generating the full film.

Think of it like releasing the full recipe, ingredients list, and step-by-step video of how a cake was baked — except the cake is a two-hour film and the ingredients include digital versions of real actors' faces and voices. The project used AI to generate every frame, every line of dialogue, and every sound, while still getting legal permission to use celebrity likenesses. All 473,600 assets were then released under an open-source license so others can inspect, remix, or learn from the exact files that built the movie.

This matters because movie-making has always required huge teams, expensive equipment, and years of work. Now one person with the right AI tools can create something that looks and sounds like a traditional film. For students interested in storytelling or art, it means the barrier to making something big just dropped dramatically. The open release also lets anyone examine how the AI handled lighting, reflections, and character consistency across the entire runtime.

You could be the next person who turns a wild idea into a full movie without needing a studio budget. The creators even open-sourced everything so others can learn exactly how they did it.

Go to the project page linked from @LinusEkenstam's post on X and scroll to the assets link. Download a few small files like character models or scene backgrounds and see how they were built. You don't need any special software to start looking at them. Source: x.com

Explain Like I'm 14

You know how when you finish a group project at school, your teacher sometimes adds a small note at the bottom with everyone's names so no one can claim they did all the work alone?

Now picture that same idea, but the note is written in a secret code that only computers can read. Every time Claude writes something, it weaves two tiny hidden marks into the text itself. One mark lives inside the actual words so it survives if someone copies the text into another document. The second mark uses an open standard called C2PA that travels with image or video files and records exactly which AI made them.

These marks don't change how the text reads or looks to you, but they let websites and apps check later whether something came from Claude. It's like adding a tiny invisible stamp that says "made by Anthropic" without anyone noticing unless they're looking for it. The system started rolling out worldwide this month and will be built into every new Claude model from day one. Older models are getting the same treatment over the next few weeks. Anthropic also plans to release free detection tools so anyone can verify the marks on their own.

The policy covers every way people access Claude, including the main website, the API, and third-party platforms like AWS or Google Cloud. Because the marks are added at the model level, they appear automatically no matter how the output is created. This approach gives a consistent way to trace AI text even after light editing, though the company notes the marks may not survive heavy changes. Source: the-decoder.com

Cool Stuff & Try This

Space tourism posters that look straight out of a 1990s sci-fi book Seedance 2.5 just turned the same simple prompt into glowing, optimistic scenes of people traveling to space like it's a normal vacation. The reflections on shiny surfaces and the bright, hopeful colors feel exactly like those old posters that predicted we'd all be flying to the moon by now. The model also handled complex lighting and mirror-like surfaces with surprising accuracy in the generated images.

It's a fun way to see how far image generators have come at making scenes feel alive and detailed. Anyone who likes making art or just daydreaming about the future should try it.

Go to any chat interface that supports Seedance 2.5 (or similar image models) and type: "optimistic 1990s-style space tourism poster showing families boarding a sleek shuttle to Mars, bright colors, reflections on helmets." Hit generate a few times and notice how the lighting and mood stay consistent. Source: x.com

Real-time news that ChatGPT and Claude can actually read Alchemiq connects directly to live news sources so ChatGPT and Claude can pull fresh information instead of only using what they were trained on months ago. The tool works by feeding current articles into the chat session, letting the models answer questions with up-to-date details.

This turns the chatbots into better research helpers for school projects or just staying up to date. You can ask questions about today's headlines and get answers that include the newest articles.

Open ChatGPT or Claude and try asking: "What's the latest on [any topic you're curious about] using real-time search?" The tool will pull current results right into the conversation. Source: Google News

Quick Bits

Singapore university makes AI class required for every new student Starting this year, every freshman at the National University of Singapore will take a compulsory AI module and get free ChatGPT access. It's one of the first big schools to treat AI literacy the same way they treat writing or math. Source: Google News

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

Forwarded this email? Subscribe here — it's free.