OpenAI has launched Sora 2, its advanced AI video generation model, with a controversial copyright policy that requires rights holders to opt out if they do not want their material used in generated videos. Announced on September 29, 2025, and detailed in a Wall Street Journal report, this approach means copyrighted characters, styles, and works will be included by default unless creators or studios explicitly request exclusion. The policy applies to the new Sora 2 app—a TikTok-like social platform for sharing AI-generated clips—raising alarms among Hollywood studios, talent agencies, and artists over unauthorized use of intellectual property.
For filmmakers, AI enthusiasts, and legal experts, Sora 2’s opt-out mechanism flips traditional consent models, placing the burden on creators to monitor and block misuse. While OpenAI emphasizes notifications to agencies and no generation of real public figures without permission, the lack of blanket opt-outs could lead to widespread litigation. With Sora 2 enabling 10-second videos from text prompts, the app has already gone viral internally, but its policy has ignited debates on fair use and AI ethics. Let’s unpack the policy, app details, and reactions.
The Opt-Out Policy: Default Inclusion of Copyrighted Material
Under Sora 2’s terms, OpenAI’s model will freely incorporate copyrighted elements—such as fictional characters (e.g., Mario, Pikachu) or artistic styles—unless rights holders submit specific opt-out requests via a disputes form. This case-by-case process means studios cannot blanket-exclude entire catalogs; they must identify and report individual violations after generation.
Key aspects of the policy:
- Default Participation: Copyrighted works are used in training and outputs unless opted out, contrasting opt-in models favored by artists.
- Notifications: OpenAI began alerting talent agencies and studios last week about the process, with the app launching imminently.
- Public Figures Exception: Real people (e.g., celebrities) require explicit permission for likeness use, verified via facial scans.
- Takedown Mechanism: Post-generation reports allow blocking specific content, but proactive exclusion is limited.
This mirrors OpenAI’s DALL-E image policy but escalates for video, where deepfakes amplify risks. Legal experts like UCLA’s Mark McKenna argue it “strikes me as not likely to work,” violating fair use precedents by shifting the onus to creators.
Sora 2 App: A Social Feed for AI-Generated Videos
Sora 2 powers a standalone iOS app (Android soon) resembling TikTok: Users prompt 10-second clips (e.g., “Pikachu storming Normandy”), share in a swipeable feed, and remix others’ creations. No camera roll uploads—everything is AI-generated—to focus on generative magic, with identity verification for personalized likenesses.
App highlights:
- Generation Limits: 10-second clips with improved physics, motion, and audio (voiceovers, sounds).
- Social Features: Likes, comments, remixing; “For You” feed curates based on prompts.
- Safety Measures: Watermarking, classifiers (90% deepfake detection), and opt-out notifications for likeness use.
Internal tests yielded “overwhelmingly positive” feedback, but viral demos (e.g., Sam Altman as James Bond) highlight misuse potential.
Reactions: Backlash from Creators and Legal Warnings
The policy has drawn sharp criticism:
- Hollywood Alarm: WME memo urged clients to opt out, calling it a “fatal blow” to IP control; Universal and Sony are monitoring.
- Artist Advocacy: SAG-AFTRA and EFF decry the “opt-out scam,” arguing it burdens creators and enables infringement.
- Legal Experts: McKenna warns of lawsuits under U.S. copyright law, where fair use doesn’t cover outputs; EU’s AI Act may impose stricter rules.
OpenAI’s Jason Kwon defended: “We treat likeness and copyright distinctly,” but critics see it as a legal gamble. On Reddit, users vent: “Opt-out is theft by default.”
Positive notes: Some indies praise generative tools for accessibility, if safeguards hold.
Implications: A Flashpoint for AI Copyright Wars
Sora 2’s policy could accelerate lawsuits (e.g., NYT vs. OpenAI) and regulatory scrutiny, with the EU probing opt-out systems under the AI Act. For creators, it’s a call to action—monitor outputs and file disputes; users, a reminder of AI’s double-edged sword; OpenAI, a test of its $157 billion valuation amid content wars.
Broader: It pressures rivals like Google (Veo 3) and Meta (Vibes) to adopt opt-in, potentially fragmenting AI video standards.
Conclusion: Sora 2’s Opt-Out Gamble – Innovation or Infringement?
OpenAI’s Sora 2 opt-out requirement for copyright holders is a high-stakes bet: Empowering viral AI videos while risking creator backlash and lawsuits. With the app’s TikTok-like feed launching soon, it could redefine social content—but at what cost to IP rights? As notifications roll out, Hollywood’s response will shape AI’s creative future. Reuters
