OpenAI released Sora 2, a text-to-video-and-audio model focused on physical plausibility, multi-shot controllability, and synchronized dialogue/SFX. The OpenAI team has also launched a new invite-only Sora iOS app (U.S. and Canada first) that enables social creation, remixing, and consent-controlled “cameos” for inserting a verified likeness into generated scenes.
Model capabilities
Sora 2 claims materially better world modeling (e.g., rebounds on missed shots instead of object “teleportation”), maintains state across shots for instruction-following edits, and generates native, time-aligned audio (speech, ambient, effects). These are framed as prerequisites for simulation-grade video generation rather than single-clip “best effort” synthesis.
App architecture and “cameos”
The Sora app is built around cameos: users record a short in-app video+audio to verify identity and capture likeness; cameo owners control who can use their likeness and can revoke or delete any video—including drafts—that includes them. The app is available on iOS devices and it will be expanding after the U.S./Canada rollout.
Safety posture
OpenAI’s Sora 2 documents an iterative rollout with specific launch-time restrictions and provenance controls:
- Uploads/Generations: At launch, OpenAI is restricting the use of image uploads that feature a photorealistic person and all video uploads. Sora 2 does not support video-to-video at launch, blocks text-to-video of public figures, and blocks generations that include real people except when a user has opted-in via the cameo feature. Additional classifier thresholds apply when a real person appears.
- Provenance: All outputs carry C2PA metadata and a visible moving watermark on downloads, with internal detection tools for origin assessment.
Parental controls
In parallel with Sora, OpenAI introduced parental controls integrated via ChatGPT: parents can opt teens into a non-personalized feed, manage DM permissions, and control whether continuous scroll is allowed—aligned with the Sora feed’s “creation-over-consumption” philosophy.
Access and pricing
The Sora iOS app is available to download now; access opens by invite, with Sora 2 initially free under compute-constrained caps. ChatGPT Pro users get access to an experimental Sora 2 Pro tier on sora.com (and coming to the app). API access is planned after the consumer rollout. Existing Sora 1 Turbo content remains available in user libraries.
Summary
Sora 2 pushes text-to-video toward controllable, physics-respecting, audio-synchronized generation—and OpenAI is shipping it inside an invite-only iOS app with consent-gated cameos plus C2PA metadata and visible watermarks for provenance. The initial U.S./Canada rollout prioritizes safety constraints (e.g., restrictions on public-figure depictions) while staging broader access and API plans, signaling a deliberate shift from raw capability demos to governed, production-ready media tooling.
Michal Sutter is a data science professional with a Master of Science in Data Science from the University of Padova. With a solid foundation in statistical analysis, machine learning, and data engineering, Michal excels at transforming complex datasets into actionable insights.
