OpenAI flips Sora IP rules

In partnership with

Welcome to VP Land! This Friday, I’m moderating the AI Innovations in Virtual Production panel at Infinity Festival Hollywood, where we’ll explore how AI, Gaussian splats, and virtual production are transforming filmmaking.

Join us at the event — use code IF2520 for 20% off passes.

In last week's poll, most of you agreed that launching a social app is a smart move for Sora to attract a broader user base. Check out today's poll below.

In today's edition:

OpenAI Updates Sora with IP Controls and Revenue Sharing
Mosaic Launches AI-Powered "Agentic" Video Editor
iPhone 17 Pro Genlock Gets Real-World Virtual Production Testing

Sora Updates Rightholder Controls, Revenue Sharing, and API Pricing

OpenAI is making two significant changes to Sora based on user feedback since the video generation tool's public launch, according to a blog post from CEO Sam Altman.

Key Updates

Enhanced rightsholder controls: OpenAI will switch its IP control from opt-out to opt-in, giving IP owners more granular control over how their characters can be generated, including the ability to block usage entirely
Revenue sharing program: The company plans to share video generation revenue with rightsholders whose characters are being used by creators. Altman notes that users are generating "much more than expected per user" with many videos created for small audiences
Rapid iteration planned: OpenAI expects "a very high rate of change" similar to ChatGPT's early development, with quick responses to feedback and course corrections

Altman describes rightsholders as "very excited" about what he calls "interactive fan fiction," though the exact revenue sharing model is still being developed. The company acknowledges that some unauthorized generations may slip through initially as they refine their content controls.

Prepare for the Sora floodgates to open: OpenAI also announced Sora 2 API (currently the only way to create was in the Sora app with watermarked videos), with pricing starting at $0.10 per second for Sora 2 (720×1280 or 1280×720) and scaling up to $0.50 per second for Sora 2 Pro (1024×1792 or 1792×1024)

SPONSOR MESSAGE

Don’t get SaaD. Get Rippling.

Disconnected software creates what we call the execution tax: wasted time, duplicate work, and stalled momentum. From onboarding checklists to reconciling expenses, SaaD slows every team down.

Rippling is the cure. With one system of record, you can update employee data once,and it syncs everywhere: payroll, benefits, expenses, devices, and apps.

Leaders gain real-time visibility. Teams regain lost hours. Employees get the seamless experience they deserve .

That’s why companies like Barry’s and Forterra turned to Rippling – to replace sprawl with speed and clarity.

It’s time to stop paying for inefficiency.

Don’t get SaaD. Get Rippling.

Stop SaaD in its tracks

iPhone 17 Virtual Production Workflow Testing

Virtual production specialist Ramiro Montes De Oca has been testing the iPhone 17 Pro Max with Blackmagic Design's ProDock for genlock capabilities in demanding virtual production scenarios.

His R&D setup pushed the system hard: a 360-degree Kino Flo MIMIK image-based lighting system running at 900fps, powered by Megapixel Helios projectors. The test specifically examined whether genlock actually works and if there's any drift over time.

Key findings from the testing:

Genlock works as promised - side-by-side comparisons confirmed proper sync functionality
iPhone 16 Pro Max and earlier models don't support the ProDock's genlock or timecode features at all
Rolling shutter performance - the iPhone 17 Pro sensor performs about 55% slower than a true global shutter, but "still good enough for most virtual production"
App bugs identified - the Blackmagic Camera app dropped frames from 120fps down to 88.5fps, and shutter angle controls behave inconsistently between iPhone 16 and 17 Pro models
Hardware appears solid - Montes De Oca attributes the issues to app bugs rather than hardware problems

The takeaway: This represents a significant step forward for iPhone-based professional workflows, but the software still needs refinement. For virtual production teams already invested in Blackmagic ecosystems, the iPhone 17 Pro's genlock support opens up new mobile capture possibilities that simply weren't available with previous iPhone generations.

Technical Hurdles Slow OpenAI-Ive Hardware Push

The secretive AI hardware collaboration between OpenAI and former Apple design chief Jony Ive is reportedly struggling with significant technical roadblocks, according to the Financial Times. After months of development and high-profile talent recruitment from Apple, Google, and other tech companies, the project remains stuck at the prototype stage with no working demos.

The article reports that OpenAI and Jony Ive are developing a palm-sized, screenless AI assistant slated for a 2026 release. The device reportedly includes multiple cameras, a smartphone-like form factor, an always-on design, and a continuously learning assistant memory.

Key development challenges:

Cloud dependency issues - The underlying AI capabilities still require cloud computing, with efforts to enable on-device intelligence reportedly stymied
No finalized technical specs - Despite the star power of its leadership, there's still no confirmed processor architecture, hardware partners, or final form factor
Missing timeline and pricing - No clarity on go-to-market strategy or when (or if) the device might actually ship
Extreme secrecy - Industry observers note that even among AI insiders, there have been no leaks about the device's real-world capabilities or unique selling points

Mosaic Launches AI Video Editor Beta

A new AI-powered video editor called Mosaic has launched in public beta, positioning itself as an "agentic" alternative to traditional timeline-based editors. The platform is live at mosaic.so with no waitlist, according to a detailed announcement thread from the company.

Seven core features define the platform:

Interactive canvas interface - Users connect tiles in a visual workspace to build editing workflows, create parallel edits from raw footage, and A/B test different prompts and models
Timeline editor with XML export - Frame-level adjustments with version history tracking, plus export capabilities for DaVinci Resolve, Final Cut Pro, and Premiere Pro
Visual intelligence system - AI analysis of video concepts, emotions, actions, speech, lighting, and shot types, with the ability to trace the agent's decision-making process during edits
Automated content workflows - Integration with Google Drive, Dropbox, and YouTube to trigger editing workflows, plus social media scheduling and platform-optimized formatting
Integrated generative AI tools - Up to 64-second video generation with character consistency, plus Runway ML, Google Veo 3, Kling AI, Deepgram transcription, and ElevenLabs audio tools
Motion graphics and asset generation - Effects like zoom-ins, Ken Burns, and color correction, plus a 30,000-asset stock library with semantic search
API access - Programmatic workflow execution with real-time progress updates via webhooks

The approach: Rather than adding chat interfaces to traditional non-linear editors, Mosaic claims to have redesigned video editing workflows from first principles based on user feedback.

Worth noting: While the feature list is comprehensive, the real test will be how well the AI actually understands video content and whether the canvas interface proves more efficient than established editing workflows for professional work.

CoPilot Virtual Production breaks down the XGRIDS Lixel L2 Pro, showcasing how this powerful SLAM (simultaneous localization and mapping) scanner transforms environment creation and integration in Unreal Engine 5.

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

🕺 Seedance Pro adds first and last frame support.

🗣️ Director Todd Rohal details how generative AI tools shaped key visuals in ‘F— My Son!’ following questions from industry peers.

🍌 Nano Banana debuts new feature set: users can now specify aspect ratios.

🎤 Taylor Swift is using AI-generated videos to market her new album, but it looks like some posts have been deleted after backlash.

▶️ Open-source model Ovi launches on fal as a Veo 3 alternative, delivering synchronized video and audio generation from both text and images.

Addy and Joey explore OpenAI’s Sora 2 — testing its likeness generation, Cameos feature, and opt-out IP policy while debating its impact on filmmakers and the gap between free and pro versions.

Read the show notes or watch the full episode.

Watch/Listen & Subscribe

Spotify | Apple Podcast | YouTube