Sora update #1

In one sentence

OpenAI is moving toward a model of "interactive fan fiction" where rights holders retain granular control over their characters while sharing in the revenue generated by user-created content.

Why read this (and who it's for)

This update addresses the friction between AI-driven creativity and intellectual property (IP) rights. It is essential for content creators, rights holders, and AI product managers who need to understand how the industry is moving toward a "permission-based" ecosystem. After reading this, you will understand the specific mechanisms OpenAI plans to use to balance user freedom with owner protections.

The big idea

The central thesis of this update is that the only way to navigate the complexities of AI-generated content is through rapid, real-world iteration. Altman argues that because the technology is moving so fast, "theory" is no longer sufficient; the company must learn from actual user behavior, rights holder feedback, and market reactions in real-time.

The logic follows a three-step progression:

1. Agency: Rights holders must have granular control (not just a binary "yes/no") over how their characters are used.
2. Incentives: A sustainable model requires a way to monetize "small audience" content, which likely involves revenue sharing.
3. Velocity: Because the landscape is shifting so quickly, the company must adopt a "high rate of change" (similar to the early ChatGPT era) to fix missteps as they occur.

Key ideas

Granular Rights Control

What it is — Instead of a simple opt-in for likeness, rights holders will have specific controls over how their characters are generated. Why it matters — It shifts the power dynamic from the AI platform to the IP owner, allowing them to decide exactly how their characters can be used (or not used at all). In the book — Altman notes that while different people will try different approaches, the goal is to let rights holders decide how to proceed.

Interactive Fan Fiction

What it is — A new category of engagement where users create content based on existing characters. Why it matters — This creates a new value proposition for rights holders: instead of just "protecting" their IP, they can participate in a new kind of engagement that may be "even more valuable" than direct revenue. In the book — Altman highlights the "remarkable creative output" of Japanese fans to illustrate the deep connection between users and content.

Revenue Sharing for Small Audiences

What it is — A plan to share revenue with rights holders who allow their characters to be generated by users. Why it matters — Many users generate videos for "very small audiences," making traditional ad-based or viral-based monetization difficult. Revenue sharing provides a way to monetize the "long tail" of content. In the book — Altman admits the exact model will require "trial and error" but emphasizes that the plan is to start "very soon."

High-Velocity Iteration

What it is — A commitment to a "high rate of change" similar to the early days of ChatGPT. Why it matters — In a nascent market, perfect decisions are impossible. The priority is to identify "missteps" and "fix them very quickly" rather than waiting for a perfect, polished release. In the book — Altman explicitly states that they will be taking feedback and iterating on different approaches in Sora specifically to inform broader product decisions.

The "Laboratory" Approach

What it is — Using Sora as a testing ground for new features and policies before applying them to the rest of the company's products. Why it matters — It allows for lower-risk experimentation on complex social and legal issues (like IP rights) before they become standard across the entire AI ecosystem. In the book — Altman mentions that they plan to do their iteration on different approaches in Sora, then "apply it consistently across our products."

Frameworks & tools

The Iterative Feedback Loop

To manage the rapid evolution of AI products, Altman suggests a cycle of constant adjustment:

1. Launch: Release a product (like Sora) to gather real-world data.
2. Listen: Collect feedback from users, rights holders, and "interested groups."
3. Identify Missteps: Acknowledge where the current stack is failing (e.g., "edge cases of generations that get through that shouldn't").
4. Rapid Fix: Implement changes quickly, acknowledging that the rate of change will be high.

Memorable examples & stories

The Japanese Connection Altman highlights the "remarkable creative output of Japan" to show how deeply connected users can become with specific content. Lesson: Community engagement is a powerful driver of content value.

Small Audience Generation The observation that "a lot of videos are being generated for very small audiences" highlights a unique challenge for AI monetization. Lesson: Success in AI generation isn't just about "going viral"; it's about finding value in the "long tail" of private or niche creation.

The ChatGPT Benchmark Altman compares the current pace of Sora's development to the "early days of ChatGPT." Lesson: Expect a "high rate of change" and frequent updates as the technology matures.

Key quotes

"We of course spent a lot of time discussing this before launch, but now that we have a product out we can do more than just theorize." Why it matters: It signals a shift from academic/theoretical safety to practical, data-driven deployment.

"Interactive fan fiction... [will] accrue a lot of value to them." Why it matters: It redefines the relationship between IP owners and AI as a collaborative opportunity.

"We will make some good decisions and some missteps, but we will take feedback and try to fix the missteps very quickly." Why it matters: It sets the expectation for a "beta" style of development where the user and the company learn together.

How to apply it

This week

• Audit your IP: If you are a creator, identify which parts of your work you would want "granular control" over (e.g., which characters, which settings, or which types of stories).
• Identify "Small Audience" Value: If you create content, think about how you could monetize a "small audience" without relying on massive viral reach.

Ongoing

• Adopt a "High Rate of Change" Mindset: When launching new projects, prioritize "launching and learning" over "perfecting before launching."
• The Feedback Loop: Establish a system for gathering feedback from "rightsholders" (or stakeholders) early in the development cycle to avoid "missteps" that require heavy correction later.

Common pitfalls

• Expecting Perfection: Readers might expect a polished, "safe" release. Altman warns that there will be "missteps" and "edge cases" that get through the system.
• One-Size-Fits-All Rights: The text warns that "different people will try very different approaches" regarding their characters. Do not assume a single permission model will work for every creator.

Connects to

• ChatGPT: The predecessor that established the "high rate of change" development model.
• Interactive Fan Fiction: A concept that bridges the gap between traditional fandom and AI-generated content.

If you remember only three things

• Rights holders will have granular control over how their characters are used in Sora.
• Revenue sharing is the primary mechanism for monetizing "small audience" AI-generated content.
• The development process will be characterized by a "high rate of change" and rapid iteration based on real-world feedback.

Internalize it

• Reflection questions

  • If I were a rights holder, what specific "granular controls" would I need to feel safe allowing my characters into an AI model?
  • How can I apply the "high rate of change" philosophy to my own current projects?
  • What does "interactive fan fiction" mean for the future of my specific niche?

• Flashcards Q: What is the primary goal of the new "granular control" for rights holders? A: To allow them to specify how their characters can be used (including not at all).

Q: How does Altman describe the current pace of Sora's development? A: A "high rate of change" similar to the early days of ChatGPT.

Q: Why is revenue sharing important for video generation? A: Because many videos are being generated for "very small audiences," and a revenue share helps monetize that content.

Q: What is the "Laboratory" approach for Sora? A: Iterating on different approaches in Sora first, then applying those decisions consistently across other products.

Q: What does Altman mean by "interactive fan fiction"? A: A new kind of engagement where users generate content based on existing characters, which can accrue value for rights holders.

Fidelity note

Confidence is high. This summary is based strictly on the provided blog post. The summary covers the two primary changes (Rights Control and Revenue Sharing), the "high rate of change" philosophy, and the specific mention of Japanese content. It does not include any technical specifications of the Sora model itself, as those were not in the provided text.